Worked on patents for high cardinality data and open source contributor to Apache Pinot, Spark, Gluten, Velox. Creating AI solutions for cost efficiency in big data.

• Designed and developed Uber's next-gen Data Observability stack handling over 2.5M+ compute jobs daily.
• Developed novel AI and distributed solutions for big data compute and storage; created features for Uber's Spark and Hadoop stack serving 100K+ QPS and 20K+ data pipelines.

Edge Computing Lab - Large Language Models, researching on parameter efficient fine tuning and adapting LLMs to domain specific usecases.

• Built large language models for computer architecture usecases, data collection of QuArch dataset which represents 50+ years of computer architecture knowledge. (Paper in progress)
• Contributed to open-source frameworks - Axolotl and Eval Harness; fine-tuned models like Llama, Mistral, Gemma, etc.

MIT Media Lab - Research in deep learning models for signal (EEG) data. Created SOTA models for classification of mental states.

• Artificial Intelligence for Brain Computer Interfaces (BCIs). Supporting ease of communication for patients suffering from neurological disorders such as Amyotrophic Lateral Sclerosis (ALS).
• Classification and detection of mental states using highly temporal and spatial EEG data. (Published 2 papers)

Research in distributed systems, cloud computing and creating cost efficient solutions for big data.

• Developing on SkyShift, a compatibility layer over resource managers like Kubernetes and Slurm, addressing scheduling problems in distributed systems.
• Created Spark jobs and load balancers, deploying them for resource manager agnostic deployments.

Worked on SGLang with researchers from Stanford University and UC Berkeley (fast-serving of large language models).

• Created custom memory pools for offloading KV cache from GPU to CPU, increasing serving throughput and reducing memory requirements for LLMs.
• Improving the throughput and serving capabilities making LLMs more efficient in distributed settings.

The Defence Research and Development Organisation is an agency of the Government of India, for military's research and development, researched on language modelling.

• Created Artificial Intelligence based systems such as Natural Language driven chatbots.
• Integrated these with the main website and backend databases such as PostgreSQL.

Brown Eagle Inc., builds next generation e-commerce platforms, based on adaptive technology with advance personalized communications and recommendations.

• Created customized Web Scraping systems designed to crawl target data source websites.
• Integrated the fully automated Data Acquisition Layer with the Internal Systems and dashboards.

KPMG International Cooperative is a multinational professional services network, and one of the Big Four accounting organizations based in Amstelveen, the Netherlands.

• Analysed and cleaned over 40,000 customers' data and worked on multiple datasets.
• Performed mutivariate and predictive analysis on the data, hence gathering deep insights and presenting them.

Decodr Technologies empower the young generation with future technology skills.They provide training and assessment programs which are designed by industry experts making it relevant for the students. They are based in New Delhi, India

• Created Machine Learning and Artificial Intelligence based projects such as Handwritten digit identification, Web Scrapers and Object Detection using algorithms like CNNs on datasets like CIFAR and MNIST.
• Research about the existing products based on similar lines in the industry.

An organisation which works on different modules related to a startup and have in-depth knowledge of growing and expanding start-ups. Primary aim is to help startups grow on different social media platforms. (Paper in progress)

• Created web-scrapers to gather and analyse data from professional websites such as LinkedIn.
• Fixing website vulnerabilities to bots and crawlers by integrating techniques like reCAPTCHA into the website.

Coding Ninjas is one of the largest online tech education company in India, focusing on courses on C++, Java, Python, Android, Machine Learning, Data science, WebDev, interview prep, tech aptitude etc.

• Debugging codes on a day to day basis and troubleshooting errors.
• Mentoring Students, monitoring their performance and taking daily doubts.

Tagore International School, Vasant Vihar, New Delhi, India

AISSCE/CBSE (Class XII) March 2017 - March 2018