Deepak Silaych
Education
B.Tech, Indian Institute of Technology (IIT) Bombay (Expected 2026)
Internships
• LiorTutor | Software Developer Intern (Dec 2024)
– Designed and deployed a microservices architecture using Next.js and FastAPI, reducing frontend latency by 70%
– Implemented a dynamic game-loading system with Redis caching and WebSocket, enabling real-time content delivery and
– Built a configurable LLM-based prompting system, allowing tutors to edit educational games via natural language inputs
– Streamlined content delivery with a scalable artifact management system, boosting delivery efficiency by 60%
• DraftAi | Software Developer Intern (July 2024)
– Engineered secure document authentication system using Aadhaar API, reducing verification time by 80%
– Developed intelligent PDF parsing library using NLP and OCR, achieving 95% accuracy in form field detection
– Built responsive document editor interface using Next.js and TailwindCSS, improving mobile engagement by 65%
– Implemented real-time collaboration features using WebSocket and Redis, supporting 100+ concurrent editors
• Healthka | Web Developer Intern (Aug – Sep 2023)
– Architected dashboard using React Query and Redux, reducing data fetch times by 60% and handling 10k+ daily records
– Implemented REST API gateway using Node.js and Express, achieving 99.9% uptime and <100ms response times
– Migrated from static to React.js architecture, resulting in 3x faster page loads and 40% increase in user engagement
Key Projects
• FinSearch: Financial Document Analysis System | Github (Dec 2024)
Achieved 1st Position at InterIIT Tech Meet 13.0 for developing advanced RAG pipeline for financial document analysis
– Engineered comprehensive RAG pipeline using LangGraph framework for analyzing 10-K filings, incorporating custom nodes
for document grading, reranking, and metadata extraction
– Implemented sophisticated question decomposition and answer generation system with GPT-4 and OpenAI embeddings
– Developed custom vector store client with timeout mechanisms using Pathway, optimizing retrieval operations
– Created evaluation pipeline using LLM-based evaluators and EntityPrecision metrics for system benchmarking
• Mumbai Flood Warning System | mumbaiflood.in (Apr 2023 – Jul 2024)
Led technical development in 20+ member team for city-wide flood monitoring platform serving 10,000+ weekly users
– Architected and deployed full-stack solution using React JS (web), React Native (mobile), hosted on VPS
– Engineered robust backend using Django, Celery, and Redis for automated data processing and alerts
– Developed ML prediction system combining GFS weather data with historical station-level measurements
– Implemented real-time flood reporting using Twitter API, NLP sentiment analysis, and Geopy location tracking
– Designed scalable PostgreSQL-API pipeline to overcome firewall restrictions and handle growing user base
• Decentralized Trading Platform | Video | Github (Dec 2023)
Achieved 1st Position in InterIIT Tech Meet 12.0 for development of decentralized trading platform on Aptos blockchain
– Led the integration of Econia protocol, ensuring a secure foundation for the innovative trading platform
– Led seamless integration with petra wallet to enhance user security and accessibility over Aptos blockchain
– Implemented market data & transaction APIs for real-time and streamlined financial transactions using FastAPI
• IITB SSO Integration Layer | LIVE | Github (Oct 2024 – Present)
Developing secure middleware to extend IITB’s Single Sign-On system beyond institutional domains
– Architecting middleware using Node.js to enable external domain access to IITB SSO while maintaining security
– Implementing comprehensive security with OAuth 2.0, JWT tokens, origin validation, and IP whitelisting
– Developed secure callback URL validation and rate-limiting mechanisms to protect against potential exploitation
• Smart Surveillance System | Github (Dec 2023)
Achieved 7th rank out of 50+ teams at Urban Futurism Competition, Techfest 2023
– Enhanced low-quality video with generative AI models, improving detail, which reduced MSE 0.08 to 0.02
– Implemented object and face recognition for identifying and tracking individuals using YOLO
– Ensured privacy-preserving surveillance through techniques like anonymizing faces (OpenCV) on public feed
• Real-Time Sign Language Recognizer | Video | Github (May 2023 – Jul 2023)
– Developed sign language recognizer using CNN model using TensorFlow & Keras achieving 97% accuracy
– Applied OpenCV to integrate webcam input, optimizing data processing for enhanced real-time performance
– Utilized MediaPipe for hand tracking and pose estimation, enabling hand detection and gesture recognition
• Real-Time CodeLab Platform (Sept 2023)
– Built interactive coding platform using WebSocket and Docker for hosting Foundation Programme’s programming contest
– Implemented real-time code execution and collaboration features supporting 300+ concurrent users with <100ms latency
– Developed secure sandbox environment using Docker containers for running untrusted code with resource limitations
– Created real-time leaderboard and analytics dashboard using Redis pub/sub and Socket.io, handling 1000+ submissions/hour
