Jan 2025 – Jul 2025 · 6 mos · Hybrid

• Developed a scalable RAG pipeline using Pinecone and FastAPI to power search and retrieval. Scraped and transformed data from Shopify stores for seamless ingestion and storage.
• Built end-to-end analytics solutions for KPIs like CTR and customer attrition, enabling clients to boost sales. Designed a custom pipeline with BigQuery and Google Pub/Sub to manage large-scale data processing and storage in GCS buckets.
• Deployed open-source models (InfiniteYOU, LLaMA-7B) on VMs, and optimised GPU utilisation by writing custom CUDA kernels.
• Created a wrapper over Vertex AI Media Studio to assist clients in AI-based content generation, competitive analysis, and social media strategy management.
• Optimised SQL queries to reduce runtime and significantly improve API latency.
• Designed an internal debugging module that reduced customer issue resolution time by 70%, streamlining technical support.
• Scaled social media moderation agents to efficiently handle 10,000+ queries, orchestrating automated workflows for client operations.

Aug 2024 – Jan 2025 · 5 mos · Hybrid

• Working as a part of the early team, working on social media moderation AI agents that scaled brand sales by 35% than general using automated replies and moderation.
• I worked on debugging the pipeline for monitoring and Prometheus monitoring. Integrated VertexAI and GCP services in the product and gained hands-on experience. Worked with Facebook and Instagram integration for their services and learned about event handling architecture.

Feb 2024 – Jul 2024 · 5 mos · Mumbai, Maharashtra, India · Remote

• Worked on ZERON’s Compliance Automation Tool
• Contributed to the development of ZERON's compliance assistant, designed to streamline and automate security compliance checks across organisational workflows. The tool helps users evaluate adherence to frameworks like ISO 27001, SOC 2, and NIST.
• Fine-tuned open-source language models (using HuggingFace Transformers) to classify compliance documents and queries with 95% accuracy, significantly improving the system’s response quality.
• Implemented advanced Retrieval-Augmented Generation (RAG) techniques, including GraphRAG, to improve contextual understanding and traceability of generated responses.
• Integrated the pipeline using LangChain and Python, enabling modular, production-ready components for LLM orchestration.
• Worked closely with compliance datasets to enhance the system’s ability to generate audit-ready responses based on policy documents and internal guidelines.
• Tech Stack: Python, Transformers (HuggingFace), LangChain, GraphRAG, REST APIs, PyTorch

Jun 2023 – Dec 2023 · 6 mos · Cayman Islands · Remote

• Worked on generative AI and computer vision features for production deployment
• Developed and deployed a pose transfer model within the production web application, enabling dynamic and realistic human pose transformation across images.
• Worked on image-to-image generation using Stable Diffusion, enabling creative visual outputs for features like virtual try-ons and AI-powered personalisation.
• Built an automated pipeline for image cloud storage and retrieval, leveraging AWS Lambda and Firehose, which significantly reduced computing time and improved system scalability.
• Designed and implemented SQL-based analytics and data logging pipelines to monitor model performance and user interaction data.
• Collaborated on end-to-end computer vision solutions, from pre-processing and model training to API integration and deployment for real-time inference.
• Contributed to the development of innovative CV features intended to improve user engagement and enhance app functionality.
• Tech Stack: Python, OpenCV, PyTorch, Stable Diffusion, AWS Lambda, AWS Firehose, Cloud Storage (S3/GCS), SQL, FastAPI/Flask

Feb 2023 – Mar 2023 · 1 mo · India · Remote

• Worked on creating front -end for the mobile app using react -native and added other functionalities as well.
• The project was successfully completed and my work has been deployed in the app .
• Also troubleshooted and corrected existing code and made it more efficient.

Jan 2023 – Mar 2023 · 2 mos · Kolkata · Remote

• Worked on ML and computer Vision projects under this apprenticeship.
• Was amongst top 15 contributors out of the pool of 2000 contributors.

Dec 2022 – Jan 2023 · 1 mo · Kolkata, West Bengal, India · Remote

• Workedon data retrieval and processing and NLP problems Under professor Dwaipayan Roy.
• Created a data set extractor it will extract data set mentioned in the research paper and make user aware of the occurrences.
• Tech Stack = Python, Grobid, NLTK