Feb 2025 – Aug 2025 · 6 mos · Bengaluru · Hybrid

• I spearhead AI Team at Delta Airlines for Quality Engineering, bringing self-healing ML and natural-language tooling to every stage of the software-delivery pipeline. From automated testing to data exploration, so teams ship features faster and with greater reliability.
• 1. Cut regression time from days to hours. Built a GenAI-powered automation suite that now covers 98% of test cases and gives developers instant feedback.
• 2. Made our ML stack self-healing. Added dynamic debugging and automatic rollback logic that keeps pipelines running and trimmed test runtime by 80%.
• 3. Turned plain-English questions into live dashboards. Shipped a Text2SQL LLM chatbot that lets any teammate pull data and charts without writing code.
• 4. Keeps answers fresh, every time. Orchestrated RAG pipelines on AWS Bedrock that fetch the latest data before each chatbot response, ensuring decisions rely on real-time numbers.
• 5. Halved test-data prep effort. Deployed a GenAI “Test Data Assistant” called DataBot that spins up realistic datasets from a short prompt, freeing QA engineers for higher-value work.
• 6. Put GenAI in every developer’s IDE. Packaged core capabilities into a lightweight VS Code SDK so engineers can generate data or run ML backed tests with one command.
• 7. Stress tested future features safely. Pioneered synthetic image, audio, and video data generation to expand coverage without touching real customer content.
• 8. Solve the next hard problem, on repeat. Continually tackle model-debugging and scaling challenges through rapid experiments and data-driven rollouts across the airline.
• 9. Project transitioned to US teams after org re-structuring.

Apr 2023 – Dec 2024 · 1 yr 8 mos · Bengaluru, Karnataka, India

• As an AI Engineer on Deloitte USI's AI Craft Team, I drive innovation and operational efficiency across AI/ML projects. Recently, I led a proof of concept (POC) for a GenAI application, implementing a Retrieval-Augmented Generation (RAG) system using LangChain for LLM orchestration, Chroma vector storage for embeddings, and GPT-3 for human-like response generation. By integrating Tavily Search, I ensured real-time, up-to-date knowledge, facilitating seamless interactions between clients and teams, boosting productivity, and reducing costs.
• My expertise includes fine-tuning large language models (LLMs) using LoRA and optimizing RAG pipelines with chunking, reranking, and embedding techniques for enhanced accuracy. I am skilled in machine learning libraries like Scikit-Learn, TensorFlow, and PyTorch for model building and training, with strong experience in data preprocessing, feature engineering, and model evaluation.
• In addition to AI/ML, I have designed Airflow-based data pipelines for real-time processing using Spark and Postgres, which I leveraged to automate manufacturing dashboards for a major IT client. This blend of AI/ML expertise and data engineering ensures efficient model development and deployment across varied projects.

Sep 2020 – Mar 2023 · 2 yrs 6 mos · Bengaluru, Karnataka, India

• As an Analyst on the Data Science Team at Deloitte USI, I worked on several impactful AI/ML projects. I spearheaded the development of a cross-client performance evaluation system, incorporating machine learning algorithms and leveraging Pandas and NumPy for data cleaning, transformation, and aggregation. I created dynamic visualizations using Plotly and Seaborn, and deployed them via Flask on the backend and Angular on the frontend, ensuring a responsive and user-friendly interface using HTML, CSS, and Bootstrap. This system enabled clients to make data-driven decisions and enhanced overall business outcomes.
• Additionally, I developed an error classification model using XGBoost with a 96.7% accuracy rate. I applied NLP techniques like TF-IDF vectorization to preprocess textual data and experimented with various algorithms to ensure optimal performance. I designed and implemented an automated ML pipeline, deployed on AWS SageMaker, allowing continuous improvement through periodic model retraining. This solution reduced computation time by 10x compared to manual processes and significantly improved classification efficiency.
• My role showcased proficiency in full-stack development, data visualization, and end-to-end machine learning, as well as expertise in cloud deployment and NLP. I was recognized with multiple awards for my technical contributions and delivery of high-impact solutions.

Apr 2019 – Apr 2020 · 1 yr

• Responsible for directing the ideology of the society and leading a group of individuals to collectively improve.

Apr 2018 – Apr 2019 · 1 yr

• Responsible for handling and improving the inflow of funds for the society.

Jul 2017 – Dec 2019 · 2 yrs 5 mos · Ranchi Area, India

• Responsible for managing the documents and the website of our NGO