Vacancies & applications
Source documents are linked below. Applications are not attendance or qualified-candidate counts.
The Ranks page now uses only the supplied official BCI provisional result annexures as its main dataset: published overall merit and four-decimal marks, roll/name search, category/gender calculated positions, printed subcategory views, PDF-page sources, stable subset positions and CSV. Positions are calculated inside the complete selected annexure (or category/group before name search), preserving published tie order. Search and pagination never renumber category positions. All 122 TSP rows also occur in the main 3,481-row list; no double counting or automatic TSP exclusion is applied.
Horizontal views read the annexure's printed EXM, SP, WID, DIV and PwD subtype strings; no third-party flag is injected. Blank source fields and missing Sahariya claims are not inferred. These positions are descriptive—not official reservation-allocation or final selection ranks. The previous conditional BCI/SCI model remains in a separate closed-by-default sandbox, loaded only when opened; its allocation switch is independently OFF. Unverified SCI data never feeds the official main explorer.
The provisional approximately 2× shortlist is now the public analysis baseline. Result Insights uses only the supplied BCI PDF-derived annexures: category/gender composition, marks bands, descriptive statistics, merit/marks curve, score counts and published-rank spotlight. All 122 TSP entries also appear in the 3,481-row main annexure; people are not double counted across lists. These visuals are not final-selection or 1× forecasts.
Model Dashboard, Deep Analysis, Estimator, Bias Check, survey Dataset and submission pages now show an archive notice, with navigation links removed. Records, numerical engines and historical comparisons are retained; no candidate/settings deletion or mutation is involved. The new all-category rank explorer supports source-flag groups with derived subset positions and conditional scenario outcomes; Sahariya identity is missing, not inferred. Dedicated EXM/PwD/Sports final allocation remains unavailable.
Shala Darpan school-level verification is still incomplete: no official report has been retrieved, so all imported school rows remain verification pending. A failed public-portal fetch, directory self-consistency or a source URL alone cannot authenticate staffing counts.
Official provisional results, published merit and calculated category positions—plus the assumptions and limitations of the separate survey models.
The Home dashboard now shows the notice’s approximately 2× scrutiny/DV shortlist cutoffs. The notice does not publish an official 1× cutoff or final appointment list. Scrutiny form dates in the notice are 12–21 November 2026; check RSSB for corrections and extensions.
The Results page searches 3,481 Annexure-1 records and 122 Annexure-2 records. All 122 TSP-annexure rolls overlap Annexure-1, so there are 3,481 unique rolls—not 3,603 different candidates. Annexure-1 TSP flags are preserved, not interpreted as an eligibility exclusion.
Published merit is copied from each annexure. Category position and category/gender position are computed within that same list, retaining published order, including equal-mark rows. They are not official category ranks, UR-allocation ranks, selection ranks or appointment guarantees. GEN-category position concerns GEN-labelled candidates only; UR/open merit includes all categories. Search and gender filters never recompute positions.
The saved 474-record model comparison is frozen and separately labelled; the model dashboard, Analysis, Estimator and Bias Check remain live hypothetical tools, not official-result calculations. Rank scenarios and † qualification floors must not be presented as official cutoffs or retrospectively tuned predictions. Existing official record tables below retain their original source scope; the new notice states 3,629 advertised posts (2,936 Non-TSP, 693 TSP), which is not silently substituted into older seat/backlog models.
Searchable records preserve official four-decimal marks, subcategory text and source page. Parents’ names are omitted. Original supplied PDFs and SHA-256 fingerprints are on Results; the separately supplied SC rankwise analysis is not treated as a board-issued category allocation list. No result records are merged into the submitted-score database.
Search official results, sources and cutoffs →The reservation audit identified incorrect UR paper-eligibility assumptions, tag-based merit exclusions, an unexplained seat deduction and an area-wide/category quota mismatch. Corrected models can differ from earlier numeric outputs. Coverage defaults, editable multipliers and historical saved snapshots are not adjusted to reach a preferred cutoff.
Official result records, published cutoffs and PDF files remain unchanged. Numerical tests establish only the behaviours they exercise; no test pass certifies all recruitment rules or candidate documents.
Read calculation coverage and outstanding limits →Published BCI rank/marks remain available. Rule-complete selection cannot be determined without candidate-wise non-fee concessions, verified EXM/PwD/Sports rosters, backlog-year history and area/department preferences. SCI’s 574 anonymous records and 291 overlap links remain third-party data.
The incomplete-data sandbox is OFF by default and is not a rules-certified allocator. Tags no longer remove candidates from ordinary merit or automatically deduct one seat each. Unsupported default56 PwD deduction was removed, not reinterpreted as no PwD reservations. BCI printed women/widow/divorce counts and same-category fallback are used in the simplified sandbox. SCI subquota details remain unresolved. Category/backlog lifecycle is separate from a final allocation, and missing preferences cannot be invented.
Sample/form/bias engines exclude demonstrably relaxed paper scores from UR assumptions; missing concession history is still explicitly unknown. Historical saved model comparisons stay frozen and are not recalibrated against published results.
Calculation audit, corrections & unresolved data →The user-supplied independent directory snapshot is now available as a native, read-only school post-status table with district, block, cadre and vacant/working filters, search and CSV export. External directory links and the iframe have been removed. Employee names are not published.
Zero school-level official records were verified. Shala Darpan sanctioned/working/vacant counts and the effective date have not been retrieved. Every imported row remains verification pending; internal consistency checks are not government certification.
These school-post figures are not RSSB 2026 advertised seats or a counselling vacancy matrix. They do not feed cutoffs, category positions, prediction models or candidate records. Dated official school-ID/cadre matching is still required.
Open School Post Status →Inclusive evidence: known paper failures are excluded; missing paper marks stay unverified. These defaults describe a conditional scenario, not a verified prediction.
Inspect the default calculation →Source documents are linked below. Applications are not attendance or qualified-candidate counts.
Approved does not mean board-verified. Unknown paper-wise eligibility stays explicitly unverified.
Coverage, migration and reporting assumptions can change the result. They are not official facts.
Changing the model is not the same as getting more evidence. Compare results only after checking their scope.
Combines official seats, application-share population allocation and usable score evidence. The 1× and fixed-UR 2× use the same population model; no category selects whichever model gives a higher score.
Open the model category cards →Retains application-based score-shape weights, but each reserved category’s population is anchored to its own usable sample ÷ coverage. Select it in Analysis or Estimator. Other older charts/tools retain their labelled sample methods.
Inspect the previous model →Changes reporting assumptions around the previous sample-coverage reference. Its former-Home reference card is pinned to model defaults; it is not a matched-control measure of reporting bias.
Explore assumptions and reporting curves →The survey model dashboard and comparison tools distinguish these outcomes.
The target has sufficient observed support under the selected assumptions. It is still conditional—not a calibrated promise of the official cutoff.
Under the selected coverage and migration assumptions, the implied eligible pool falls below the target. The base qualification minimum is shown as a labelled floor scenario—not an independently estimated rank cutoff.
The result changes materially with UR migration treatment or lies near the sample edge. The asterisk is not a confidence score, and the method spread is not a predicted-result interval.
Thin, missing or category-gated evidence does not establish a defensible number. A lowest submitted score is an observation; it is not automatically a population bound or a cutoff forecast.
Floor scenarios use the applicable base list minimum only when the documented lower-tail exhaustion checks pass. They do not solve relaxed-claim eligibility or joint horizontal allocation. MBC’s observed sample minimum, for example, need not equal its qualification minimum.
Site submissions require approval before entering the public dataset. Imported survey totals are another source of self-reported evidence—not an official result or independent marks verification.
Hidden, invalid, inconsistent-total and duplicate-ID records are excluded from comparison inputs. A known failure in either paper makes an entry ineligible; total-only entries remain eligibility-unverified.
The forms scenario divides usable Non-TSP evidence by assumed coverage, then allocates that implied population by category application shares while retaining the existing gender weights. UR/open merit uses all categories, not GEN alone, and needs at least 5 usable entries in every required category. TSP has no official application-area mix and is not forms-calibrated.
For a reserved list, 1× targets list seats plus reported-qualified UR migration. The default 2× doubles the list seats but retains that same 1× UR migration adjustment. It does not double marks, qualifying minima or the migration adjustment, and is not proof of the board’s calling rule.
Scores are ordered high to low. In the forms scenario, weights express implied population positions, and the target is interpolated within that weighted distribution. The previous sample-coverage model instead maps a target to weighted sample positions. Neither method establishes a supported rank estimate from sparse evidence alone.
A supported rank result, a qualification-floor scenario and sparse/gated evidence are different states. The label and reasoning are part of the result—not optional fine print.
N = n ÷ CN[c] = N × A[c] ÷ An = usable Non-TSP entries; C = assumed overall reporting coverage; A[c] = category application forms; A = six-category application total. N and N[c] are implied potentially eligible populations, not observed pass counts.
1× target = S[c] + M₁[c]2× target = 2 × S[c] + M₁[c]S[c] = official list seats with the chosen backlog setting; M₁[c] = estimated reported-qualified migration above the 1× UR boundary. UR itself uses all categories and has no reserved-list migration addition.
Illustration only: these round numbers are not any real category, dataset snapshot or seat allocation.
The weighted score distribution is still needed to translate ranks 140 and 240 into marks. Seats/forms alone cannot produce a cutoff, and the 15%/20% assumptions here are not measured response or pass rates.
The model sandbox uses a forms-calibrated Non-TSP scenario: total implied pool = usable Non-TSP scores ÷ assumed coverage; each category pool = total pool × its application share. Existing category/gender score weights are retained. All-area application shares are an unverified proxy for eligible Non-TSP shares—not measured area, attendance or pass distributions. The default 2× retains the 1× UR adjustment.
The previous sample-coverage comparison remains selectable in Analysis and Estimator. For that earlier method:
weighted sample target = (list-seat target + estimated UR migration) × coverageThe default coverage is 15%, with backlog included. It is not a measured survey response rate. Coverage/backlog changes apply to the page and its shared link; they do not establish a new official baseline.
The default forms-calibrated 1× and fixed-UR 2× use the same population model. No category picks the higher of different models. The recomputed convention finds a 2× UR boundary and recalculates the reserved candidates migrating to open merit. The fixed-UR convention (the current default) retains the 1× migration adjustment while doubling the reserved-list target. Neither treatment is established as the board’s official calling rule by this analysis.
Only paper-verified scores meeting unrelaxed 40/100 minima can contribute to model migration. Known non-fee concessions disqualify UR eligibility; missing concession history remains an explicit assumption, not confirmed migration. Candidates above the UR score with unknown paper marks are not silently treated as verified.
Inspect the full 1× / 2× comparison →Where a tool shows a bootstrap range, it repeatedly resamples the submitted evidence under that tool’s labelled settings. Some comparisons deliberately omit bootstrap calculations. Resampling does not measure non-response bias, prove the coverage assumption or guarantee that the official result falls inside the displayed range.
Higher scorers may be more willing to report their marks. A uniform response-rate assumption cannot identify that bias from the submitted rows alone.
Change the within-cell score distribution while preserving the existing category/gender totals and implied population. They isolate score-shape sensitivity.
Assume absolute reporting probabilities by score. These also change the implied population. “15% reporting at score 100” is not the same as 15% average coverage.
The curves are hypothetical, not measured confidence functions. No scenario is designated “best” or selected to match a desired cutoff. Bias Check retains the previous sample-coverage reference so its audited reporting-bias profiles remain comparable. It is not the forms-calibrated model sandbox. Its controls are page-local and do not change the homepage calculation or saved calculation preferences.
The reporting curve shows relative propensity for fixed-pool profiles and assumed percentages for absolute-rate profiles—not confidence. Rank-to-rank differences are calculated only when both results are supported rank estimates; floor transitions are labelled, not subtracted as measured cutoff changes. CSV exports identify the separate Home-default settings.
Explore reporting-bias scenarios →Available as an optional reporting scenario in Bias Check and an expandable comparison in Deep Analysis. The model sandbox keeps its approved forms-calibrated 1× / fixed-UR 2× default. This experiment is not an official rule, a calibrated prediction or an accuracy upgrade.
| Score band | Default multiplier | Implied reporting rate |
|---|---|---|
| <70 marks | 10× | 10.00% · assumed |
| ≥70 and <80 marks | 10× | 10.00% · assumed |
| ≥80 and <90 marks | 10× | 10.00% · assumed |
| ≥90 and <100 marks | 10× | 10.00% · assumed |
| ≥100 and <110 marks | 7× | 14.29% · assumed |
| ≥110 and <120 marks | 5× | 20.00% · assumed |
| ≥120 and <130 marks | 3× | 33.33% · assumed |
| ≥130 marks | 3× | 33.33% · assumed |
The original community message used 120+:3×, 110–120:5×, 100–110:7× and under100:10×, with ambiguous boundary overlap. Our explicit interpretation puts exact 100, 110 and 120 in the higher band. The requested finer grid splits 70–130 into 10-mark intervals, plus separate below70 and130+ tails. Defaults repeat the original rule within these bands; no independently measured rates were supplied for the finer splits. Every finite valid score belongs to exactly one band. Both tools expose a seat-to-rank table using official list seats for the selected area and backlog: 1× seats and 2× seats, the applicable UR migration addition and the resulting weighted rank target. Seat counts are never multiplied by reporting factors.
scenario weight = existing demographic weight × band multiplierassumed reporting rate = 1 ÷ multiplierThe score is unchanged. Unlike the forms-calibrated model sandbox, this experiment does not force a forms-allocated population total. There is no second division by15% and no forms-weight overlay. Changing a multiplier changes both the implied pool and potentially its score shape.
The control is the previous sample-coverage model at uniform15%, using the same category, area, evidence, backlog and migration settings. Deep Analysis coverage does not rescale this separate experiment. Rank-to-rank differences alone can be numeric; floor or unsupported transitions remain qualitative. A difference from Home is not solely a reporting-bias effect.
All eight band multipliers are editable from1 to100×, including fractional values; this bounds assumed reporting rates between1% and100%. These are interface safety limits, not empirically justified ranges. Edits are drafts until Apply multipliers succeeds. Blank, non-finite or out-of-range values do not replace the applied result. Restore community defaults resets only this scenario. A reversal of the higher-scorer-reporting assumption is allowed but flagged.
Shared URLs carry applied multipliers, not unsaved drafts. Deep Analysis also carries whether its panel is open; its Bias Check link carries the exact selected controls. Community CSV exports include every applied score band and multiplier. Invalid multiplier links fall back to the documented defaults with a visible warning when the experiment is shown.
Known paper failures remain excluded; missing paper marks remain unverified. Only reported-qualified entries above the chosen UR boundary contribute to migration. Fixed-UR 2× retains 1× migration; recomputed uses its own2× UR boundary. Existing thin-sample, UR support, TSP and floor gates remain in force. Weighted pool counts are not verified passed-candidate counts. Sahariya is not a separately calibrated forecast. The community message mixes SCI-labelled discussion with a BCI-seat table: its posted cutoff claims are not imported as BCI evidence.
Try editable community multipliers → · Open the Deep Analysis comparison →The site’s rule set uses 40/100 in each paper for GEN/EWS/OBC/MBC, and 35/100 for SC/ST. Both papers must qualify; an aggregate of 80 or 70 alone does not prove that.
Example: 60 + 20 = 80 overall, but the second paper fails the ordinary 40-mark minimum.
Both known papers clearing the applicable threshold counts as reported paper-qualified evidence. Missing paper marks stay unverified. Known failures are excluded even when the total looks high.
DV/PwBD and ex-serviceman claims use the existing relaxed per-paper rule where applicable. A horizontal category label does not by itself prove entitlement; official rules and verification govern.
UR/open merit is not a GEN-only list: eligible candidates from other categories can compete. In TSP, EWS/OBC/MBC do not have separate lists in this model and are directed to UR. Sparse TSP evidence is not filled with an invented cutoff in the evidence comparison.
Women, widow/divorcee, disability and ex-serviceman reservations overlap the vertical lists; their counts must not be added as extra seats or distinct candidates. The site does not establish a full joint allocation across all these claims. Some historical TSP-discount tools, where explicitly labelled, are assumptions—not a substitute for a supported observed boundary.
Women can compete through open merit as well as applicable horizontal reservation. The women’s diagnostics use their stated female evidence, list and quota assumptions; they are not a verified final allocation simulation. When the sample cannot reach the target, its observed minimum is not a reliable upper bound or forecast.
Printed women’s seat splits are distinguished from assumptions; TSP backlog splits may use a flagged 30% assumption where a separate printed split is absent.
View the women’s evidence and states →These values come from the project’s documented source records—not from a fitted score curve.
User-provided RSSB Advt. 07/2026 detailed notification (RajKaj Ref. 22397996), pages 5–8 — every district row checked against the board's printed TOTAL row. Refer to the latest board document for amendments or the final selection rules.
View category, district and horizontal seat tables →Figures are the reply's FEMALE + MALE online-application columns summed; the reply publishes no official TSP / Non-TSP split, so area shares stay explicit assumptions. Online applications are not attendance, paper-qualified candidates or distinct eligible merit-pool counts.
Open the reply image · bindu 01 (opens in a new tab)| Category | Female | Male | Total |
|---|---|---|---|
| GEN | 9,582 | 21,011 | 30,593 |
| EWS | 3,480 | 11,090 | 14,570 |
| OBC | 9,912 | 34,661 | 44,573 |
| MBC | 484 | 2,779 | 3,263 |
| SC | 3,383 | 14,483 | 17,866 |
| ST | 2,412 | 9,693 | 12,105 |
| Saharia | 4 | 28 | 32 |
| Total incl. Saharia | 29,257 | 93,745 | 1,23,002 |
Six-category total: 1,22,970. PH 1,121 and EX 579 are horizontal counts already included in the totals; they are not added again. The reply does not provide an official TSP/Non-TSP application split.
| Category | Female | Male | Total |
|---|---|---|---|
| GEN | 3,271 | 5,017 | 8,288 |
| EWS | 749 | 1,761 | 2,510 |
| MBC | 50 | 241 | 291 |
| OBC | 1,966 | 5,008 | 6,974 |
| SC | 577 | 1,681 | 2,258 |
| ST | 140 | 474 | 614 |
| SAHARIA | 1 | 0 | 1 |
| Total | 6,754 | 14,182 | 20,936 |
| Department | Post | Non-TSP | TSP | Total |
|---|---|---|---|---|
| Secondary Education | Senior Computer Instructor | 268 | 32 | 300 |
| Secondary Education | Basic Computer Instructor | 2,736 | 679 | 3,415 |
| Sanskrit Education | Senior Computer Instructor | 20 | 2 | 22 |
| Sanskrit Education | Basic Computer Instructor | 200 | 14 | 214 |
| All posts | Basic + Senior | 3,224 | 727 | 3,951 |
Cross-check: department rows match the recorded combined total. Basic-only department totals match the seat-table totals.
Advt. 07/2026, RajKaj Ref. 22397996; seat tables on pages 5–8. The site keeps a local copy so the tables remain inspectable.
Board-hosted PDF (opens in a new tab)Local PDF copy (opens in a new tab)SHA-256: 77f18354a059d3822ee46527e7fe728487759424813f81b9a03c4fa15ae59b2e
This identifies the stored document. It is not independent verification of every model assumption or a check for later board amendments.
The supplied reply records female/male online applications and the department vacancy table. It does not supply a measured survey coverage rate or an official area-wise qualified-candidate count.
Applications reply image (opens in a new tab)Department vacancy image (opens in a new tab)Approved site entries and imported JK Byte survey evidence underpin the score sample. JK Byte totals alone cannot verify per-paper eligibility. Neither source is a random sample.
Inspect the public dataset →JK Byte public leaderboard (opens in a new tab)Historical final cutoffs, written/DV-stage cutoffs, attendance reporting and media expectations answer different questions. They are not interchangeable, and none measures this sample’s current reporting probability. Expected-cutoff articles are contextual references—not official results or calibration targets.
No. It corrects category representation under the hypothesis that all-area application shares approximate eligible Non-TSP shares. Different categories may have different area membership, attendance, paper qualification and willingness to report. Those differences are not measured by the application table.
The model still needs an overall coverage assumption. More inputs or a higher cutoff do not prove better prediction accuracy. The site does not choose the maximum across models to produce a desired answer.
Home now shows official approximately2× scrutiny cutoffs. The separate model sandbox uses forms-calibrated 1× and fixed-UR 2× by default. Bias scenarios use their labelled previous-model reference and selected reporting curve. Its former-Home reference card always uses 15% coverage, inclusive evidence, backlog included and fixed UR; only category/area follow the Bias page. A model-sandbox link using different coverage/backlog may also differ.
Match category, area, population model, eligibility evidence, reporting/coverage assumptions, backlog, UR migration convention and dataset snapshot before interpreting a difference.
The application reply supplies no official TSP/Non-TSP category mix. TSP keeps its existing unweighted evidence gates rather than borrowing a Non-TSP application allocation. Sahariya is a separate seat list outside the six-category calibration denominator and needs separately identifiable eligible score evidence. Neither is filled with an invented cutoff.
The forms scenario gates inconsistent coverage settings if the implied six-category eligible pool exceeds that application-form universe. Missing category support, low effective sample size and unavailable migration also limit what can be shown. Passing these checks does not establish that the remaining assumptions are true.
No. These are conditional score/rank scenarios. Official eligibility, category/area entitlement, calling rules, document verification, vacancies and final allocation govern selection.
The eligible score distribution, demographic weights and conditional UR migration can change as approved observations change. Coverage, backlog, evidence and migration choices also matter. Compare the same dataset, area, list, stage and assumptions before interpreting a change.
Not necessarily. A larger opt-in sample can still over-represent confident or higher-scoring candidates. Independent response-rate evidence or a sampling frame is needed to calibrate that bias; Bias Check only explores explicit assumptions.
Approved rows may appear in the public dataset. Review that page before contributing and avoid unnecessary personal information. Public visitors cannot edit or delete approved entries; request corrections through the site administrator. Approval is an access/moderation step, not an RSSB marks audit.
The footer displays a published four-hour analytics snapshot, not a live event counter. New-tab visits count as visits; refresh/retry deduplication prevents the same event being counted repeatedly. Another tab in the same browser is not automatically a new unique visitor. Browser/device estimates are not a count of identified people.
Community acknowledgement: thanks to the Computer instructor pass students (opens in a new tab) group for discussion, feedback and outreach. Community participation is not board endorsement or independent verification.