My focus is on the full AI lifecycle: from R&D in multimodal and agentic systems to building production-ready, optimized pipelines that deliver real-world impact.

Hackathon & Competition Wins:

• 🥇 1st Place, AI Hiring Show by Rabbitt AI: Led winning team (500+) for an LLM financial assistant.
• 🥉 Winner, Hire-A-Thon (Geek Room & InvoLead): Top 3 out of 800+ for advanced speech AI solutions.
• 🥉 Winner, SkyHack 2.0 (United Airlines): Recognized among 1600+ candidates from top institutes.
• 🏅 5th Place, Data Engineering Summit 2025: E-commerce Product Rating Prediction.

Research & Kaggle:

• 🏆 Top 10 (International), LT-EDI@LDK 2025: Misogyny Meme Detection shared task.
• 🏆 Top 10, FIRE 2025 DravidianCodeMix: Offensive Language Identification shared task.
• 💡 Kaggle Top 6%, NIFTY50 Options Volatility Prediction: Outperformed nearly 2,000 participants.
• 💡 Kaggle 17th/600+, SHL Grammar Scoring: Developed a BERT-based NLP system for audio transcript evaluation.

Professional Recognition:

• 🥇 Keploy API Fellow ‘25: Top 5% of 18,500+ global applicants.
• 🛡️ UGC NET Qualified (June 2024): Top 6% for Assistant Professor & Ph.D. admission.
• ✈️ AFCAT Qualified (Aug 2024): Recommended for SSB.
• 💻 400+ DSA Problems Solved across LeetCode and other platforms.

Airavata LLM Quantization Pipeline | Python, PyTorch, Transformers, FastAPI, Docker

Mission: To make a 6.87B parameter LLM practical for production by shrinking its size and increasing its speed.

• Impact: Achieved 4x memory reduction and 2-4x faster inference using INT8/INT4/GPTQ techniques.
• Features: Built a production-ready FastAPI backend and a complete Dockerized benchmarking suite to measure latency, throughput, and resource utilization.

Enhanced LinkedIn Sourcing Agent | Python, LLM Integration, FastAPI, Async Programming

Mission: To build a multi-source agentic workflow that automates candidate discovery and scoring.

• Impact: Enabled 4x faster candidate sourcing with a more accurate, AI-powered technical credibility score.
• Features: Integrated multiple LLMs (Gemini, Groq) with A/B testing, and validated data across LinkedIn, GitHub, and academic sites.

Voices-Reimagined (Speech-to-Speech AI Pipeline) | Wav2Vec2, Pyannote, SpeechBrain

Mission: To create a real-time system that understands the full context of a conversation, not just the words.

• Impact: Won a Top 3 prize at a national-level hackathon for its advanced capabilities.
• Features: Integrated multilingual transcription, speaker diarization, emotion detection, and summarization into a single, real-time workflow.

RAGBot for DUCS | FAISS, Sentence Transformers, Google Gemini

Mission: To build an academic chatbot to provide accurate, real-time answers for 500+ university department users.

• Impact: Reduced query latency and improved answer accuracy for departmental Q&A.
• Features: Implemented a Retrieval-Augmented Generation pipeline using FAISS for efficient knowledge retrieval and Sentence Transformers for semantic search.

• Data Science Intern (GenAI), Involead: Led GenAI research (Knowledge Distillation, RAG); published internal whitepaper; built/deployed scalable ML pipelines for clients using AWS, MongoDB, and Docker.
• Machine Learning Intern, IBM (CSRBOX): Built SVR models for academic prediction; applied data analytics for SDG-focused projects.
• Data Analyst, Dusker AI: Automated data ETL and reporting; scaled SQL analytics on live education data to deliver insights to product leads.
• Data Science Intern, CodSoft: Delivered data cleaning, preprocessing, and modeling for real-world data science projects.

• Sentence Transformers, UKPLab: Identified and helped resolve a critical JSON serialization bug in the library, enhancing stability for global developers. The fix was successfully merged.
• Shared Tasks (Peer-Reviewed):
  • LT-EDI@LDK 2025: Top 10 globally in Misogyny Meme Detection (multimodal deep learning).
  • FIRE 2025 DravidianCodeMix: Top 10 in offensive language detection for code-mixed Indian languages.
  • CLMIR 2025: Designed and evaluated crosslingual math information retrieval systems.

• Machine Learning with Python (freeCodeCamp)
• IAB Digital Marketing & Media (Google)
• IoT (Stanford)
• Arduino ATMega (MoE-IIC/DU)
• Cybersecurity (Cisco)

Open to:

• AI/ML, GenAI, and NLP engineering roles, research fellowships, and open-source projects.
• Collaborations at the intersection of LLMs, RAG, Agentic AI, and social impact.

Contact:

• LinkedIn • Portfolio • LeetCode • Twitter

"Building AI that is robust, explainable, and transformative—bringing advanced technology to every language and user worldwide."