Jun 2025 – Jan 2026 · 7 mos · New Delhi · On-site

• Developed an NL2SQL chatbot on large PostgreSQL datasets using Ollama models, advanced prompt engineering,
• and local fine-tuning. Integrated the system with predictive ML pipelines to generate pattern-based insights and
• automated query responses.
• Created an OCR-driven document automation system combining fine-tuned LLMs with large annotated datasets
• to automate text extraction, document classification, and end-to-end processing at scale.

Jul 2024 – Jun 2025 · 11 mos · Chennai, Tamil Nadu, India · Remote

• Real-Time Trading & Analytics Infrastructure :
• Engineered low-latency trading platforms for equities/crypto with streaming pipelines, automated signal generation, and ML-driven predictive modeling; built modular backtesting engines and drag-and-drop strategy builders for collaborative experimentation.
• Generative AI & Autonomous Content Intelligence :
• Architected LLM-powered platforms (Groq LLaMA3-70B, custom transformers) for product narratives, semantic audio-visual sync, and content matching; developed scalable scraping + async LLM workflows and automated Blender pipelines, boosting efficiency by 60–70%.
• ML-Ready Systems & Intelligent Automation :
• Designed end-to-end ML infrastructures with dynamic filtering, predictive signals, and reinforcement learning; combined time-series analytics, 50+ indicators, and semantic embeddings for adaptive decision-making in finance & media AI.

Oct 2022 – Jul 2024 · 1 yr 9 mos · Chennai, Tamil Nadu, India · Remote

• Deployed a large-scale fuel optimization algorithm across 300+ vessels, reducing operational overhead through efficient Python, ECL, and Bash automation.
• Built scalable PySpark pipelines for real-time maritime analytics, enabling faster insights from large datasets.
• Engineered secure Flask backends with JWT authentication, integrated with Nginx, AWS Lambda, and S3 to support high-availability marine systems.
• Developed serverless workflows using Lambda, EventBridge, and SNS for live JSON monitoring, automated error alerts, and vessel geolocation tracking.
• Migrated compute-intensive PDF generation from AWS Lambda to AWS Fargate, significantly improving scalability and cost efficiency.
• Optimized complex SQL workflows, improving reporting and analytics performance by 30% across cross-functional teams.
• Integrated Mistral and LLaMA models using LangChain to build multilingual NLP decision-support systems.
• Built Zero Shot Visualizer, a web app that transforms natural language queries into interactive visual insights using LangChain, SerpAPI, and Matplotlib.
• Automated EC2 instance provisioning with Lambda and launch templates for seamless machine learning model deployment.
• Created event-driven Python pipelines for real-time data processing, alerting, and maritime operations monitoring.

Nov 2021 – May 2022 · 6 mos

• Led end-to-end data preparation workflows including cleaning, feature engineering, imputation, and normalization to enhance model accuracy and robustness.
• Built a high-performing artificial neural network for precise customer attrition prediction, driving actionable business insights.
• Applied diverse supervised and unsupervised learning techniques such as logistic regression, linear regression, k-nearest neighbors, decision trees, and convolutional neural networks for advanced data classification and analysis.
• Leveraged NumPy, pandas, and Scikit-learn to develop scalable preprocessing pipelines ensuring data integrity and model readiness.