Dec 2024 – May 2025 · 5 mos · New York, United States · Remote

• Specialized in training and refining large language models (LLMs) using Reinforcement Learning from Human Feedback (RLHF) methodologies to enhance model performance and alignment with user intent
• Conducted rigorous evaluation of LLM outputs using structured rubrics to assess factual accuracy, coherence, verbosity, and adherence to instructions
• Performed side-by-side comparisons of model responses, identifying strengths and weaknesses to determine superior outputs and provide actionable feedback for iterative improvement
• Evaluated AI-generated responses from user prompts and API calls, ensuring accuracy, relevance, and overall quality
• Conducted multi-dimensional evaluations of AI outputs, focusing on coherence, factual accuracy, and alignment with user needs

Apr 2024 – Dec 2024 · 8 mos · San Francisco, California, United States · Remote

May 2023 – Apr 2024 · 11 mos · Remote

• Scaled learning operations from 100 to 1,000+ concurrent students while maintaining 95% satisfaction rate through data-driven program optimization and quality improvements
• Built and managed a network of 25+ industry experts and technical trainers to deliver specialized tracks in Data Science, Cloud Computing, and DevOps
• Developed and implemented bilingual quality control framework for content delivery, ensuring consistent learning outcomes across English and regional language programs
• Forged strategic partnerships with 10+ leading tech companies for curriculum validation, industry alignment, and graduate placement opportunities
• Optimized training budget allocation across multiple learning programs, resulting in 30% cost reduction while maintaining 95% client satisfaction

Jan 2021 – Feb 2023 · 2 yrs 1 mo · New York, United States

• Spearheaded a strategic Computer Vision project, in collaboration with the CEO, to revolutionize warehouse management, tracking, and compliance for a $5B+ client, impacting 5,000+ locations
• Designed an innovative stereo vision system for COVID-19 physical distancing compliance using a hybrid approach of StereoSGBM for initial estimation, enhanced by Transformer-based cost volume refinement on a Raspberry Pi 4, achieving <3% depth error across 50+ jobsites
• Engineered a real-time, proprietary worker tracking system with uniquely patterned, high-contrast HSV tags, optimized for robust detection under varying lighting conditions using OpenCV and DETR
• Created an auto-annotation pipeline with Mask2Former and Labelme, reducing dataset preparation time by 65% and shortening model iteration cycles by 45%
• Productionized deep learning workflows with Docker on Kubernetes (GKE) and HAProxy load balancing to handle 8,000+ API requests/hour, reducing inference costs by 30%
• Established real-time performance monitoring with Prometheus and Grafana, enhancing overall delivery efficiency by 22% through data-driven decision making and stakeholder alignment

Dec 2019 – Dec 2020 · 1 yr · New York, United States

• Led cross-functional teams of 30+ global collaborators on 3 AI for social good challenges, establishing project roadmaps, sprint planning, and deliverable timelines
• Spearheaded a computer vision project for disaster response using satellite imagery analysis, resulting in a model with 87% accuracy in identifying damaged infrastructure
• Implemented Agile methodologies with weekly sprints and daily standups across distributed teams, improving project completion rates by 35%
• Created comprehensive product requirement documents (PRDs) and technical specifications for NLP solutions focused on misinformation detection
• Facilitated collaboration between data scientists, engineers, and domain experts through Slack, Trello, and GitHub, ensuring alignment with stakeholder requirements
• Developed a water scarcity prediction system using time-series forecasting and GIS data integration, presented to NGO partners for implementation in drought-prone regions
• Conducted user research and converted findings into actionable feature priorities, increasing model adoption rates by 40% among partner organizations
• Organized bi-weekly demo sessions showcasing incremental progress to stakeholders, gathering feedback and iteratively refining solutions
• Managed project documentation and knowledge sharing through Notion and GitHub wikis, ensuring project sustainability beyond challenge periods
• Built a dashboard for visualizing project KPIs and model performance metrics using Streamlit and Plotly, facilitating data-driven decision making

Nov 2018 – Jun 2021 · 2 yrs 7 mos · Remote

• Directed a team of 5 machine learning engineers to deliver 12 Computer Vision based MicroSaaS solutions to clients in contract roles over 9 months
• Advocated the team's research outcomes and in-house data science tools in 10 conferences and 25+ hackathons across the globe
• Crafted curriculum for partner firms to teach Data Structures, DevOPs and Open Source principles to students aged 14-18
• Supervised a team of 7 machine learning researchers with the co-founder which precipitated in a cumulative output of 30 research papers from the organization
• Launched multiple Chatbots using Google DialogFlow and Rasa for partner firms for better client management
• Acquired funding of 5000$ USD for internal projects by drafting research grants and advocating the impact of the team
• Provided paid mentoring sessions for prospective Master's thesis students who intended to work at the intersection of Computer Vision and Medical Imaging
• Presided over 20 workshops on research paper writing, scientific surveying methods, and implementation of transformer models

Mar 2018 – Sep 2019 · 1 yr 6 mos · London, England, United Kingdom

• Developed an image classification pipeline using TensorFlow to identify make and model of knee implants resulting in achievement of 96.72% validation accuracy
• Coordinated collection of 50000 radiographs from hospitals in India and United Kingdom by creation and maintenance of a non-relational database utilizing Firebase, thereby diminishing staff time for the process by 65%
• Investigated impact of modern augmentation techniques including CutMix and MixUp for reducing the generalization of the deep learning model and leading to an overall increase of 0.25 in the F-1 score
• Managed a team of 7 researchers and developers to launch an end to end system hosted using Heroku and StreamLit for enabling beta testing by orthopedic surgeons in 10 hospitals located across India and United Kingdom

Aug 2017 – Jan 2018 · 5 mos · Remote

• Devised image segmentation models utilizing U-Net with PyTorch for separation of probable carcinogenic regions from the image for screening of cervical cancer with an intersection over union score of 0.9
• Deployed a chatbot leveraging RASA and DialogFlow for assisting 180 patients in the collaborating medical facilities with an easily accessible platform to clarify diagnosis or therapy related queries instantly
• Trained a three member team of medical students to analyze various histopathological images with an interactive dashboard having opencv in it's core and eventually generate reports to automate the initial phase of diagnosis

May 2017 – Jun 2017 · 1 mo · Chennai,India

• Designed low-cost prototyping board using Eagle software to cut down cost of resources involved in robotics bootcamps for underprivileged children to 80 USD from the previous requirement of 130 USD per kit
• Analysed pre-existing research to reduce model size to 45 kilobytes for integration in ARM Cortex based microcontrollers used in achieving home and office automation related tasks for over 100 clients
• Accelerated curriculum creation for instructing high school seniors in applied machine learning by working with 2 embedded design engineers and 3 web developers