Software Engineer — Developer Tools, AI Systems, Full Stack
I build tools and product systems that make complex software easier to inspect, verify, and operate safely.
- Portfolio: subhajitpradhan.vercel.app
- Resume: one-page PDF
- LinkedIn: linkedin.com/in/subhajitlucky
- Email: subhajitpradhan310@gmail.com
- Location: Odisha, India
| Project | What it proves | Inspect |
|---|---|---|
| Codebase Doctor | Model-independent, read-only codebase auditing CLI for developers and coding agents, with structured findings, SARIF, changed-scope analysis, MCP, secrets/dependency checks, and database/RLS analysis. | npm · case study |
| RLS Doctor | Read-only Postgres/Supabase RLS auditing CLI with CI and Agent Skill workflows. | npm · case study |
| SmritiFlow | Repository-memory CLI that generates structured handoff artifacts for coding agents. | npm · case study |
| Tarka Sabha | Multi-agent debate workflow with provider routing, persona state, and encrypted credentials. | live · case study |
| CSCosmos | Typed computer-science learning hub with search, domain/topic routes, and microsite links. | live · case study |
| CampusHelper | Full-stack campus lost-and-found flow with auth, uploads, search, and claims. | live · case study |
| IntentPay | AI-assisted Web3 transaction prototype with natural-language intent, a review-before-execution wallet boundary, and spending guardrails. | case study |
I am a Full Stack Software Developer at Giakaa Capital. My work spans Next.js, TypeScript, PostgreSQL, backend APIs, job processing, authentication, and product features that need clear operational boundaries.
Core: TypeScript, JavaScript, Python, SQL, React, Next.js, Node.js, PostgreSQL
Engineering: Express.js, Prisma, Docker, GitHub Actions, REST APIs, authentication, testing, CI/CD
AI: OpenAI, Anthropic, Gemini, AI agents, MCP
Additional: Go, Rust, Solidity, Ethereum, ICP
- Prefer small, inspectable interfaces over decorative complexity.
- Keep credentials, data access, and irreversible actions behind explicit boundaries.
- Treat tests, CI, README files, and tradeoff notes as part of the product.
- Prefer evidence-backed diagnostics over speculative findings.
- Ship honest status labels: published CLI, live platform, or prototype.

