🎓 Recent Graduate: I completed my B.Tech in Computer Science with a specialization in Artificial Intelligence and Machine Learning from Gyan Ganga Institute of Technology and Sciences, scoring a CGPA of 8.95.
💡 I'm a passionate tech enthusiast who thrives on building innovative solutions using Deep Learning frameworks and AI. With a strong academic foundation, I've gained hands-on experience through diverse projects in AI, Computer Vision, and Reinforcement Learning.
I am currently working as GenAI developer @ TCS, where i am responsible for building Llm pipelines for reliable solutions and managing a team of fresher group of genAI developers.
- Python
- C++
- SQL (Intermediate)
- TensorFlow, Keras, PyTorch
- NumPy, Pandas, OpenCV, FAISS
- Stable baselines3, langChain_community
- Google ADK, Langgraph
- Jupyter Notebook
- VS Code
- GitHub
- Docker
- Flutter
- Autonomous Self-Healing RAG: Developed a multi-agent "Manager-Worker" architecture that uses iterative validation loops to dynamically expand semantic search scope (+2 results per cycle) when confidence thresholds aren't met, ensuring high-accuracy incident resolution.
- Governed Text-to-SQL & HILT: Engineered a secure analytics pipeline that generates SQL from metadata only—keeping raw data out of the LLM context—integrated with Human-in-the-Loop (HILT) guardrails for final solution validation and automated knowledge-base generation.
- Developed an AI model to compose original melodies using LSTMs and GRUs.
- Explored innovative architectures to refine music generation with TensorFlow and the
music21library.
- Built an automated trading agent leveraging DRL models like PPO, SAC, and TD3.
- Validated performance using historical data from Yahoo Finance, outperforming human traders.
- Utilized VGG16 and VGG19 for state-of-the-art texture synthesis through transfer learning.
- Achieved high-quality textures with benchmarks, confirming VGG-based architecture efficacy.
- Conducted a comparative study using models like Dlib, Haar Cascade, and MediaPipe.
- Evaluated accuracy, speed, and robustness across datasets for real-world applications.
- Built a QnA Rag agent which is trained on userspecific documents and provides answer to the context.
- It uses ensamble retrival mechanism along with globally recognised vectore store and embedder.
- Deep Learning: NPTEL - IIT Ropar(GLOBAL CERTIFICATE, Score:84%)
- Reinforcement Learning: NPTEL - IIT Madras (GLOBAL CERTIFICATE, Score:67%)
- Python: Cisco Network Academy
- Advanced C++: Cisco Network Academy
- AWS Cloud Foundations: AWS Academy
- Building RAG Agents with LLM: NVIDIA
- Multiple google cloud certifications on GenAi and VertexAI, can be found in my LinkedIn
"Dream big, code bigger, and leave a legacy to inspire!"