a keyless web unlocker for ai agents: reads, searches and transcribes what a naive fetch can’t. cli + mcp server + python lib, on pypi through 0.14.0.
0 api keys 3 unlock tiers 8 ai-chat providers
i ship hard, i review hard.
i’m aadarsh, a react-first engineer. i ship quickly and review harder.
now — india · remote-first · open to relocate
selected work.
four things you can click. every claim has a link. that’s the whole review process, right here.
a pr reviewer in two forms: hare bot in cursor chat, hare as a github app dogfooded on searchts. same review contract, and neither one can block a merge.
127 app reviews 25 bot reviews 0 blocking
google maps turn-by-turn on the back of a phone. the next maneuver lights up the nothing phone’s glyph matrix. no maps api key.
137 leds 12 maneuvers 2.3 mb apk
a free git review companion that sits beside your ai editor: the graph, the diffs, every worktree, refreshing live as your agent rewrites the repo under you.
worktree-first live refresh bring-your-own-agent
- also shipped / pitched in
- dooper, a self-hostable matrix inbox; every messaging feature free on your own server.
- wmux — not mine: diagnosed a main-process freeze, landed 4 merged prs.contributed
- burncard: know what your coding agents actually cost. npx burncard
- msu halls register — alumni-register prototype for 16 halls (engagement paused).
- coffeebreath + a discord voice overlay, older windows toys with no public url. ask me.
how i work.
what you get when you hire me. short version: ownership, evidence, docs, honesty. fast is fine. fast + reviewed is the point.
my assistants don’t get the last word.
you get an owner, not a ticket-taker. on a small team i’ve owned ui, api, data, deadline math, deploy and the written handover.
ai does the typing, evidence does the talking.
my review loop reads diffs, commits and ci as evidence, never instructions. findings land as comments. nothing there can block a merge.
the docs are the interface.
if agents use your tool, write for the agents: runbooks, skills and examples, not a raw fetch wrapper.
a confident wrong answer is worse than an honest error, and that is how i build.
my tools fail loud on login walls and captchas instead of returning junk. a tool that can’t say no can’t be trusted to say yes.
where i’ve been.
two products, one training contract, one degree in progress. open any row for what i built, the numbers, the stack.
software engineer and architect
complyv (fka compliance sarathi) · appson · jan–jul 2026
primary engineer on a small team (~553 of 647 commits) on an indian corporate-governance saas. governed ai writes (propose-then-confirm, the model never mutates data), a deadline engine for statutory dates and penalties, and a sha-256 append-only audit log. react 18 + node + mongodb.
- ~553 of 647 commits (~84%) on a small team. not the only developer; wrote the handover when the contract ended 31 jul 2026.
- propose-then-confirm writes: the model never mutates data. a separate authenticated endpoint re-checks auth, company access and business rules with idempotency keys.
- three-tier truth wall: due dates and penalties come only from the deadline engine. the draft linter is code, and eval cases bait number leakage on conceptual answers.
- sha-256 append-only audit log with an admin integrity verifier; multi-tenancy as an ownership-and-grant graph over cin, deny-by-default.
- ~1,400 loc deadline engine over multiple mca forms, 34 mongoose models, ~40 suites / ~2,852 tests. fixed an llp fee-slab bug and a dir-3 kyc triennial rule that would have been wrong from 2027.
junior software engineer
wordibly · appson · jun 2025–early 2026
owned tops-transcript-editor (~107–113 commits), a vite/react transcript editor for production transcription work on a us transcription platform. thinner share on the wider codecommit platform: orders, uploads, workforce dashboards.
- tops-transcript-editor (~107–113 commits): a vite/react transcript editor with find/replace and related editing flows, built for production transcription work.
- thinner share on the wider codecommit platform (tops-api, tops-new-ui, tops-old-ui): order and upload surfaces, workforce dashboards, editor workflows for pms, transcribers and qa.
- sprint delivery with no visual regressions on production-bound ui.
ai training engineer
turing (via appson) · 2025 · ~2 weeks
short contract producing meta llama training data and eval sets. hands-on llm data work.
- short contract (~2 weeks, 2025) via appson: meta llama training data and evaluation sets.
bca (honours), msu baroda — expected 2028
product space pm fellowship, top fellow — 2025–26
what people say.
verbatim, with sources. the last one has no name attached. so i attach none.
#133, which is one of the best bug reports this project has had. … You attached a debugger to the main process, took a CPU profile, and came back with 92% of self-time in readdirSync/statSync/readFileSync under walkDir/readCurrentFile, plus a before/after table showing system.identify going from timeout to 0.7 ms.
“from your own IP” is the whole insight. a proxy pool fights the bot-wall; your own IP is already through it, same reason a human’s browser doesn’t trip cloudflare. an agent acting from where you already are doesn’t need to sneak in. nice build.
fetch-time + final_url turns read output into something a reviewer can cite later. Tier/status is useful; redirect + timestamp makes it durable.
Aadarsh owned the architecture of Compliance Sarathi end to end and shipped a reliable agentic assistant under real deadline pressure. He is who you want on the hard parts of a system.
say hi.
currently wrapping the degree and shipping searchts + grove. fastest reply is email.
tell me the system and the deadline.
hi@capad.fyioutside-india roles with visa sponsorship welcome.
home stack: typescript · react · next.js · css · electron · node → deploy