| Education | B.E. Computer Engineering — Thapar Institute of Engineering & Technology, 2024–2028 |
| Based in | India |
| Works on | Durable backend systems · static analysis · native macOS · agent and evaluation tooling |
| Currently | Building uncertainty-aware fraud-model auditing in BlindSpot and extending the native macOS agent monitor AgentBar |
| Reach me | Portfolio · LinkedIn · shikhardeepgoel@gmail.com |
I build durable backend systems and developer tooling — crash-safe storage, fault-tolerant dispatch, and analyzers that produce a working exploit rather than a warning. The agent and evaluation work runs on the same stack: deterministic checks around probabilistic models.
Selected work merged or landed upstream. Every row links to the merge or landing commit.
| Project | Change | What it did |
|---|---|---|
| google/adk-python | #6419 ✅ shipped in v2.6.0 and v2.8.0 |
fix: handle Windows paths in adk eval — the eval-case parser read the colon in C:\ as a case selector. Google lands patches through Copybara, so the PR reads closed while the change sits on main as 6f6106f, authored by me. |
| numpy/numpy | #32386 ✅ merged | BUG: close duplicated file descriptor if fdopen fails — C-level fix for a descriptor leak in NumPy's error path, with regression tests handling WASM/musl. All 91 required CI checks passed. |
| apache/airflow | #71691 ✅ merged | fix: resolve deprecated imports in airflow.utils.helpers — the deprecation shim used bare __import__ on dotted paths, returning the top-level module instead of the submodule, so moved functions like render_template_as_native raised ImportError. Switched to importlib.import_module, with regression tests. Landed as bb3e71d. |
| apache/magpie | #1089 ✅ merged | fix(license-compliance-audit): handle large blobs and canonical SPDX — files over ~1 MiB return without inline content, so the audit silently mis-flagged them as non-compliant. Fetches via raw-media API, separates uninspected from violating. Closed issue #944. |
| topoteretes/cognee | #4126 ✅ merged | fix(cli): reject dry runs in API dispatch mode — --dry-run was silently ignored when --api-url was set, executing real remote operations. |
| corsairdev/corsair | #1200 ✅ merged | feat(pinecone): production-grade integration — 48 operations across four API surfaces, typed Zod schemas, dynamic host routing. Landed as f7820d6, +3,948 across 26 files, closing issue #1199. |
| risa-labs-inc/BossConsole | #324 ✅ merged | fix(plugins): serialize concurrent first loads per class name — made plugin class loading parallel-safe by serializing same-name definitions while allowing independent class names to load concurrently. Landed as 56de814. |
| risa-labs-inc/BossConsole | #325 ✅ merged | fix(plugins): stop host resource fallback after unload — prevented an unloaded plugin from resolving missing resources through the host while preserving the existing missing-result warning. Landed as 895c57e. |
| Project | What it is |
|---|---|
conveyor-job-queueGo Connect RPC Protobuf Prometheus |
Durable distributed job queue built around crash-safety: a CRC32C-checksummed, length-prefixed write-ahead log with torn-write detection, flushed with F_FULLFSYNC rather than plain fsync; group commit so N concurrent submitters share one flush; lease and fencing logic with jittered backoff and dead-lettering. 32,232 submissions/sec, 7.7 ms recovery for 100K records.The hard part: a worker that stalls for 30 seconds and comes back must not be able to acknowledge a job someone else now owns. The badges above are regenerated by CI, not typed here. |
redoscopeTypeScript Node ≥22.18 zero deps |
ReDoS static analysis that verifies its own findings. Parses the full ECMAScript regex grammar, compiles to an NFA modelling real backtracking, detects exponential and polynomial paths via product automata, then generates a witness attack string and times it inside a killable child process. Benchmarked on 627 real-world regexes: a star-height heuristic gave 10 false positives and missed 14 real vulnerabilities; this pipeline hit 0 false positives and 0 misses. The hard part: "looks risky" is not a vulnerability. The tool has to produce a concrete string that actually hangs the engine, and prove it with a measured growth curve. |
rescuerelayNext.js 15 TypeScript Leaflet Recharts |
Explainable crisis-resource coordination platform. Camps and clinics report needs, providers register available resources, and a deterministic matching engine ranks compatible pairings by priority, distance, availability and fairness — with a plain-language explanation attached to every recommendation. Offline-capable field intake for low-connectivity reporting. |
LiveEnv — private repoNext.js SQLite WebSocket IndexedDB MapLibre |
Local-first, decay-based geospatial social platform across four surfaces. Geohash-partitioned SQLite serving a hot TTL-swept table alongside durable transactional tables; paid placements settled through an append-only double-entry ledger so money never moves without minting the pin; IndexedDB offline outbox with idempotent client-ID-keyed sync. |
AgentBar — private repoSwift Xcode Keychain SQLite |
Native macOS menu bar app unifying usage tracking across Codex, Claude Code, Gemini, Cursor, OpenCode and Z.ai into one stacked bar with per-service metrics. Secure credential handling through macOS Keychain across heterogeneous sources; shipped as a signed .dmg under MIT. |
stateDiagram-v2
direction LR
[*] --> Pending: Submit, appended to WAL and flushed before ack
Pending --> Leased: Lease granted, epoch incremented
Leased --> Done: Ack carrying the current epoch
Leased --> RetryWait: Nack, or lease timeout reclaim
RetryWait --> Pending: backoff with jitter
RetryWait --> DeadLetter: retry budget exhausted
Leased --> Leased: Ack carrying a stale epoch — rejected
Done --> [*]
DeadLetter --> [*]
The epoch is the fencing token. A lease timeout and an explicit Nack decrement the same
retry budget, so a poison-pill job cannot loop forever without reaching the dead-letter queue.
| Project | What it is |
|---|---|
BlindSpot — private repoPython scikit-learn Streamlit pytest |
Uncertainty-aware auditor for fraud-model declines under censored outcomes. Freezes a verification policy before outcomes are revealed, estimates false-decline rates with known sampling propensities, and reports when the review budget is too small to support a stable claim. Replayed a 590,540-row IEEE-CIS benchmark byte-for-byte; 40 tests keep sealed labels out of the product path. |
WebMCP Incident WorkspaceWebMCP JavaScript Vercel · live demo |
Human-agent incident command room with nine browser-native tools. Agents can compare deterministic remediation futures, but stale plans are rejected and irreversible actions remain human-only; successful execution emits a retrievable decision receipt. Verified through seven real-client scenarios and 18 automated tests. |
modelgauntletNext.js Zod Ajv Vitest |
Pre-release testing platform that stress-tests structured AI tasks across open-source models and returns a deterministic SHIP / FIX / BLOCK verdict. One rule enforced end to end: AI proposes, deterministic TypeScript code decides — AI never grades AI. 23 automated tests across parsing, assertion, schema-validation and verdict boundaries. |
multimodal-evidence-reviewPython Gemini 2.5 Flash |
Multimodal damage-claim adjudication pipeline — model output validated and repaired against a strict 14-column schema with tightly constrained enums. Semantic consistency rules lifted claim-status accuracy from 70% to 85% with zero additional model calls. SHA-256 content-addressed caching, resumable batch pipeline, graceful stop on quota errors. Ranked #37 of 1,773. |
kaamtwinNext.js TypeScript |
Causal digital-twin simulator that reorders manufacturing order queues by deadline priority — 62 → 59 late deliveries in the demonstrated scenario, verified by 15 anti-hardcoding causal tests (reversed-input, capacity, duration, immutable-hash) that rule out a hardcoded result. 67 automated tests at 88%+ coverage, 8/8 Chromium e2e, hardened CSP. |
zero-token-routerPython Fireworks AI API |
Hybrid router resolving deterministic tasks locally and escalating only genuinely hard work to Fireworks-hosted models. Passed 19/19 simulated evaluations shipped as a public Linux/amd64 image under 4 GB RAM / 2 vCPU. |
AI Intern — Bharat Electronics Limited, Bengaluru · Jul – Aug 2026 Backend API integrations in FastAPI, a React/Vite analytics dashboard turning application data into usage, latency and performance views, and improvements to the project's RAG pipeline.
Remote AI Intern — Facillima · May – Jun 2026 Shipped an AI-native EdTech platform end to end in Next.js/TypeScript — technical SEO, structured data, analytics instrumentation, third-party LMS integration — and led evaluation of the company's LLM observability stack.
Certifications — Anthropic AI Fluency, MCP, Agent Skills and Subagents · Walmart Global Tech Advanced Software Engineering · Deloitte Cyber · Google Ads Apps
Languages
AI & agents — where most of my work sits
Agent runtimes & interfaces
Protocols, platforms & orchestration
Concepts: RAG · autonomous agents · multi-agent orchestration · deterministic verification · model evaluation & benchmarking · release gating · prompt engineering · LLM observability
Tools: LangSmith · Braintrust · Zod · Ajv · Vitest · agent skills & persistent memory
Machine learning & data
Concepts: ensemble methods · CNNs & Grad-CAM explainability · feature engineering · model calibration · ROC-AUC / permutation importance
Systems & infrastructure
Concepts: write-ahead logging · crash recovery & durability · fencing tokens · lease-based dispatch · group commit · static analysis / automata · local-first & offline sync
Backend, frontend & stores
Core CS: data structures & algorithms · DBMS · object-oriented programming · distributed systems · software testing (pytest, Vitest, Playwright, Ruff)
Conveyor's crash-safety and throughput figures are re-measured on every push by verify.



