Dhruv Kushwaha
Computer Science undergraduate (2027) specialising in backend engineering, with hands-on experience building REST APIs, authentication flows, and relational and document data models using Node. js, Express. js, and FastAPI. Interested in building scalable and maintainable backend systems, clean API design, and applied AI.
Strong sense of ownership — comfortable carrying a component from design through testing, integration, and deployment — with a bias toward automation and a habit of picking up unfamiliar tooling quickly. Comfortable working across cross-functional teams.
Experience
Intern
Sawai Labs
Interned at Sawai Labs, an operator-led AI product and transformation lab, working on agentic AI systems and client delivery. Joined through the firm's AI Generalist Accelerator Programme, which combined structured technical grounding with live project work.
Technical grounding: Building blocks of agentic systems in Claude Code — prompt engineering, context and memory management, reusable skill and tool design, and external service integration via the Model Context Protocol (MCP), focused on how instructions, memory, and tool access compose into reliable agent behaviour. Agent development: Built Travel Companion on the Open Claw harness, a Telegram-deployed travel assistant covering itinerary generation, budget-aware planning, and shared expense settlement, with a persistent memory layer for personalised recommendations and group-scoped data isolation.
Containerised with Docker Compose. Client delivery: Built an MVP for a client engagement, from initial requirements through to a working prototype.
Education
LNM Institute of Information Technology
B.Tech.
CGPA: 7.79 / 10
Boston Public School, Agra
12th
Percentage: 90.80 / 100
St. Francis Convent School, Agra
10th
Percentage: 96.50 / 100
Skills
JavaScriptSQLMyPythonCC++SolidityBackendBlockchainMongoDBPostgreGitHubDockerPostmanBurp-SuiteRAGHTMLREST APIExpress.jsNodeJSOOPSDBMSData Structures and AlgorithmsAgentic AIOpenClawLLM IntegrationMCP
PROJECTS
AgroChain – Decentralized Agricultural Waste & Produce Supply Chain Platform
01 Oct 2025 - 20 Nov 2025
Designed a role-based access control model across six participant types, layered over an eight-stage shipment lifecycle state machine with transition validation enforced on-chain. Developed five Solidity smart contracts– identity/KYC registry, oracle manager, ERC-721 asset tokens, ERC-20 escrow, and dispute resolution– with replay protection, signature verification, and reentrancy safeguards; automated settlement splits payment across participants on verified delivery.
Unit-tested with Foundry. Built an event-driven synchronisation layer that commits blockchain events to the application database only on transaction finality, maintaining consistency between on-chain and off-chain state.
VideoServe
Built a versioned REST API for registration, login, session refresh, credential updates, and profile management, with bcrypt password hashing and JWT access/refresh token authentication enforced through reusable middleware and cookie-based sessions. Implemented a file ingestion pipeline using Multer to handle multipart/form-data uploads– staging files to temporary local storage, streaming them to Cloudinary, and cleaning up temporary files on both success and failure paths.
Modelled users, videos, and subscriptions in Mongoose with indexed lookup fields, referential relations for watch history and channel subscriptions, and pagination on aggregation queries; standardised behaviour through a shared async handler and common error/response wrappers.
Travel Companion Agent– Conversational Travel Assistant
Built an agentic travel assistant featuring Travel DNA personalisation, itinerary generation, destination discovery, budget aware trip planning, and flight assistance. Integrated web search, scheduled workflows, and memory-driven recommendations so the assistant carries user preferences and trip context across sessions.
Added Telegram group support for collaborative trip planning, expense management, and travel coordination, with per-group data scoping; containerised for deployment with Docker Compose.
HackerSec — AI-Driven Security Code Review System
Team Size: 2
Built a multi-stage vulnerability analysis pipeline combining static analysis (Semgrep, Bandit), code property graph taint tracking via Joern, and RAG-grounded LLM reasoning over CWE/OWASP references retrieved through FAISS. Implemented asynchronous job processing with a FastAPI service backed by a Celery worker pool and Redis broker, exposing upload, status, and results endpoints so long-running scans run off the request path.
Fused signals from each analysis stage through a scikit-learn classifier with SHAP-based explanations, and generated suggested patches; packaged the full stack with Docker Compose.
EXTRA CURRICULAR ACTIVITIES
Hackathons
Participated in various hackathons, including AMD Hackathon, Smart India Hackathon (SIH), LNMHacks, Code-e-Manipal, HackRx by Bajaj, GenAI Hackathon, and Google Developer Group (GDG) events.
PERSONAL INTERESTS / HOBBIES
Interests
I enjoy exploring emerging technologies like AI and Web3. I’m passionate about creating impactful projects and learning new tools. Beyond tech, I love playing various sports to stay active, disciplined, and team-oriented.