• Scaled revenue from $0 to $100k during 6 months of operations.• Designed and developed the Translatetracks app backend using Python and FastAPI.• Implemented a transcription engine integrating Speech-to-Text APIs such as Deepgram, Gladia, and Assembly AI, along with multimedia tools like FFmpeg, Moviepy and Pydub.• Developed a versatile translation engine leveraging advanced models such as ChatGPT-4, ChatGPT-3.5, and embeddings from OpenAI and Cohere• Fine-tuned models like Llama-2-7b using LORA and Hugging Face's Parameter Efficient Fine-Tuning (PEFT) for custom functionality.• Developed predictive machine learning models to automate and optimise the dubbing pipeline.• Developed and implemented agents within the dubbing and translation pipeline using Langchain Agents and custom tools• Tech stack - Python, Pytorch, scikit-learn, FastAPI, AWS, LangChain, HuggingFace, PEFT, OpenAI, Llama, MongoDB, SQL.

• Developed and deployed backend for the application (inspired from Nikita Bier’s Gas App) utilizinig Python FastAPI, MongoDB, and PostgreSQL.• Devised innovative user acquisition strategies leveraging Instagram, optimizing engagement and conversion rates.• Successfully scaled the application's user base to six schools in Delhi, with a total of 2,500 users.

• Deployed AI-based feed monitoring system across multiple recycling plants and Material Recovery Facilities (MRFs) in India and Europe.• Designed image acquisition pipelines for real-time capture in multi-camera setups.• Selected for the NVIDIA Inception Startup program and AWS Accelerate, receiving $100K in AWS compute credits.• Trained millions of images on hundreds of object detection models (Yolo), segmentation models, and image classification models (Googlenet, Resnet, MobileNet, etc.) using Pytorch, Weights & Biases, and Tensorboard.• Developed in-house image labelling tools and systems to monitor errors, model drifts, and image searches within our datasets.• Deployed an end-to-end inference pipeline on NVIDIA Jetson for real-time monitoring, processing 100 objects per second.• Developed the backend for a monitoring dashboard using Python FastAPI and PostgreSQL.• Tech Stack- PyTorch, Flask, AWS, Weights and Biases, TensorRT, Docker, Kubernetes, AirFlow, React, MongoDB, SQL.

• Performed exploratory data analysis to identify critical trends and features influencing process efficiencies.• Conducted feature engineering and sensitivity analysis to optimize model predictions for key performance indicators.• Implemented machine learning solutions for real-time predictive analysis.• Tech Stack - Python, Scikit-Learn, SciPy, Matplotlib, and Seaborn

As the convener of the fest I was responsible for,- Planning and strategizing activities throughout the year.- Effective collaboration within different teams of the fest.- Introducing external participation and raising the footfall substantially. - Launching new events and increasing the scope of student involvement.- Successfully organizing the events.

As the creative head, my roles involved- Working cohesively with the artists and designers - Develop creatives for the events- Designing posters and other publicity content. - Also designed the official Fest T-shirt using Photoshop.

Completed a project on "Harvesting Algae using gravity driven microfluidic hydrodynamic separation of particles"