Every number on the resume, re-checked daily
Receipts
Each claim on my resume points at its public source. Every morning a job fetches the source and records whether it still says what the resume says. A claim counts only when the source says it word for word.
Last checked:
- 23 Matches
- 5 No public source
Proof strip
- No public source
27 merged PRs on a live payments platform
Company repository (Zabira Academy) is private; the count is re-run by the owner with gh search prs.
- Matches
Upstream PR jd/tenacity #668 is merged
merged = true / Source
- Matches
Upstream PR dry-python/returns #2480 is merged
merged = true / Source
- Matches
Upstream PR memgraph/gqlalchemy #390 is merged
merged = true / Source
- Matches
13.4k stars combined: jd/tenacity has at least 8,800
stargazers_count = 8836 / Source
- Matches
13.4k stars combined: dry-python/returns has at least 4,300
stargazers_count = 4388 / Source
- Matches
13.4k stars combined: memgraph/gqlalchemy has at least 300
stargazers_count = 302 / Source
- Matches
Report with DOI: FLAKE LAUNDERING
source answered with a page / Source
- Matches
Report with DOI: COLD READ
source answered with a page / Source
- Matches
Report with DOI: AUGUR
source answered with a page / Source
- Matches
Published dev tool: tiebreak on PyPI
info.name = "tiebreak-rerank" / Source
- Matches
Published dev tool: autocto on PyPI (package repo-autocto)
info.name = "repo-autocto" / Source
- Matches
Published dev tool: skilltether on npm
name = "skilltether" / Source
- Matches
Published dev tool: Docstep on the VS Code Marketplace
source answered with a page / Source
Experience
- No public source
Zabira Academy: 24 audit findings (1 critical), production incident root cause, 113 end-to-end assertions
Private company repository; source is the internship log kept by the owner.
- No public source
Pratyaya Limited (Nobero): support analytics dashboard, 292 tests, 10 of 10 totals reconciled
Private company repository with customer data; never published.
Selected work
- Matches
claimledger: a memory library's published LongMemEval recall reproduced exactly, 483 of 500
line 7 / Source
- Matches
claimledger: plain BM25 on the same questions scored 476 of 500
line 7 / Source
- Matches
claimledger: the gap is not significant (McNemar p = 0.28)
line 7 / Source
- Matches
FLAKE LAUNDERING: 229 scored coding-agent runs
line 3 / Source
- Matches
FLAKE LAUNDERING: the agent weakened a correct test in 0 of 60 runs
line 6 / Source
- Matches
FLAKE LAUNDERING: "do not change the tests" cut correct repairs from 9 of 10 to 0 of 10
line 11 / Source
- Matches
tiebreak: +0.05 nDCG@10 on NFCorpus
line 16 / Source
- Matches
tiebreak: +0.11 nDCG@10 on FiQA
line 19 / Source
- Matches
tiebreak: the README prints the small loss on SciFact
line 13 / Source
- Matches
Docstep: adds an Update Docstring command that keeps what you wrote
line 53 / Source
- No public source
Docstep: 331 unit tests on three OSes
The count comes from the test run (npm run unit_test), not from a page that states it; CI logs expire.
Education
- No public source
B.Tech IT, MGM University, CGPA 8.19
University records are not public; the marksheet is available on request.