Data Scientist II at Recruiterflow (2024-12 – Present)
- AIRA Notetaker, an LLM-powered notetaker for structured extraction and summarization for recruiter calls, reducing post-call documentation time by over 70%.
- Architected Matchmaker, a candidate-to-job ranking engine combining vector search with LLM re-ranking, improving match relevance for 10,000+ active job requisitions.
- Built multi-step agentic workflows using LangGraph for automated candidate screening, follow-up emails, and profile enrichment, eliminating 60%+ of repetitive recruiter tasks.
- Developed scalable RAG pipelines with active feedback loops, enabling semantic search across recruiter call transcripts, resumes, and job descriptions.
Data Scientist I at Tech9 (2020-01 – 2024-11)
- Founded and scaled the AI function as the company's first data scientist, defining ML strategy and aligning it with business goals across multiple verticals.
- Built an LLM-powered consultation call analysis pipeline using LangChain and GPT-3.5-turbo, deployed on AWS Step Functions; processed 20,000+ calls/month and reduced manual effort by 80%.
- Designed a serverless debit transaction classification system on AWS, using a 1-D CNN (TensorFlow) to categorize 35 transaction types with 82% accuracy, improving processing speed by 80%.
- Built and deployed end-to-end ML pipelines on AWS (Lambda, Fargate, ECR, Bedrock) and GCP (Vertex AI, Cloud Run) using Docker, MLflow, and Kubeflow.
Junior Data Scientist at Datametica Solutions Pvt. Ltd. (2018-06 – 2019-12)
- Built a social media rumour detection system for an Indian Stock Exchange using an LSTM model (TensorFlow), achieving 90% accuracy and 0.88 recall for real-time classification.
