A Software Engineering student at INSAT with strong experience and passion for AI/ML. Currently exploring advanced topics in deep learning, image processing, and data engineering.
My experience spans full-stack development, machine learning engineering, and MLOps, with hands-on involvement in AI projects ranging from computer vision to RAG systems.
I actively participate in hackathons and technical challenges, with wins at IEEE TSYP and Orange AI Hackathon. Always driven to bridge the gap between research and production by building scalable, intelligent systems.
Pursuing Software Engineering with specialization in AI/ML. Core curriculum includes:
AI & Data Science: Artificial Intelligence, Machine Learning, Deep Learning, Data Mining & Analysis
Image Processing: Image Processing, Computer Vision techniques
Databases: Database Administration, NoSQL Systems, Data Management
Backend Architecture: Software Architectures, Distributed Systems, Microservices
DevOps & Infrastructure: Docker, Kubernetes, CI/CD pipelines
Quality Assurance: Software Testing, Quality Assurance practices
Complementary coursework in numerical analysis, algorithms, system design, and computer networks.
Advanced work in AI/ML systems design. Engineered RAG-powered reporting pipelines with semantic search and structured LLM analysis. Built talent-matching and document processing systems leveraging modern AI architectures.
Deployed and architected conversational AI systems integrating LLMs with messaging platforms. Built serverless support infrastructure on AWS with LangChain chatbots, WhatsApp Business API, and automated escalation workflows.
Architected serverless LLM service using AWS Lambda and Bedrock for geospatial intelligence. Benchmarked foundation models and engineered advanced prompting strategies to automate technical report generation, improving accuracy by 25%.
Engineered time-series forecasting models (SARIMA, Prophet) for complex prediction problems. Developed RAG-based legal AI platform using LangChain and Pinecone, reducing document analysis time by 70%.
Developed full-stack subscription and payment management platform using NestJS, React, and Apollo GraphQL. Optimized data layer with Prisma ORM and PostgreSQL, reducing fetch requests by 40%. Containerized deployment with Docker.
Designed full-stack web application for payment device lifecycle management. Developed RESTful API using Spring Boot with JWT authentication, React.js frontend, and PostgreSQL database integration.
Here are some of the skills I've acquired over the years.
A selection of AI, ML, MLOps, and data engineering projects demonstrating expertise in research and production systems.
Three-stage HTR pipeline using Qwen3-VL for layout analysis and line-faithful transcription on historical manuscripts. Integrated Qwen2.5-72B refinement layer with LLM-as-a-Judge semantic evaluation framework, reducing word-level error rates across Spanish datasets.
Tech: VLMs, Qwen, OCR, LLMs, Historical Document Analysis
End-to-end platform using DeepPurpose (Morgan + CNN) for binding affinity prediction (0.808 Concordance Index) with semantic search across 23K+ compounds (sub-350ms latency). Next.js dashboard with 3Dmol.js visualization and multi-agent pipeline (Generator→Ranker→Validator).
Tech: DeepPurpose, Qdrant, Next.js, 3Dmol.js, Multi-Agent Systems
Researching and implementing context compression techniques (LongLLMLingua) to solve context rot in long-running database conversations. Architecting multi-tier memory compression pipeline with question-aware scoring and document reordering for Guepard's QweryAgent.
Tech: LLMs, Context Compression, Database Querying, RAG
Personalized knowledge management with graph-augmented retrieval (semantics+centrality+recency scoring) using Qdrant vectorDB, PostgreSQL, Redis, and Ollama. Auto-ingestion pipeline supporting PDF, Word, images (VLM captioning), audio (Whisper) with React constellation UI.
Tech: RAG, Qdrant, Ollama, React, D3.js, Multimodal AI
Developed robotic chess coach system that uses computer vision to analyze the chess board during games and predict optimal moves for tutoring elementary students. Integrated robotics control with deep learning-based board recognition.
Tech: Computer Vision, Robotics, Deep Learning, Board Analysis
Recognized for innovation and excellence in AI/ML and software engineering competitions.
1st Place Winner for an innovative technical solution combining embedded systems, signal processing, and software engineering principles.
2nd Place Winner for the Energy Optimization System for Photovoltaic (PV) Management project. Participated in a technical challenge with excellent framework combining physics-based PV simulation, ML forecasting, and predictive maintenance.
3rd Place Winner for participation in the churn challenge. Leveraged AI/ML and cloud technologies to address telecom operational challenges in a 24-hour hackathon focused on Telecom Operations & Business Innovation.
Feel free to reach out to me if you have any questions or if you'd like to collaborate on a project.