Naman Pant

AI Researcher

India4 yrs experience

AI EnabledAI ML Practitioner

Key Highlights

Expert in building end-to-end ML solutions.
Proven track record in enhancing AI system accuracy.
Strong foundation in computer vision and generative AI.

Stackforce AI infers this person is a Data Scientist specializing in AI and Machine Learning for E-commerce applications.

Contact

Skills

Core Skills

Machine LearningData Science

Other Skills

AI AgentsAPI DevelopmentAgentic FrameworkAlgorithmsAmazon S3Amazon Web Services (AWS)Application Programming Interfaces (API)Artificial Intelligence (AI)Back-End Web DevelopmentBashBenchmarkingC++Competitive ProgrammingComputer ScienceComputer Vision

About

Working at the intersection of Data Science, ML/AI, and software engineering to solve complex and problems for the past 3 years. I have built solutions in different sub-domains of AI like classical ML, Computer Vision, GenAI and NLP using a lot of different GenAI models(chatGPT, llama3, stable diffusion, CLIP). What I do: - As a data scientist and ML Engineer, I often find myself developing the end-to-end solution of a problem, from building ETL pipelines, preforming EDA (Airflow), brainstorming different approaches for model training and inferencing (MLflow), to deploying and monitoring final solution. Skills I have: Languages: Python, Pytorch, XGBoost, SKLearn, OpenCV Frameworks and Databases: Flask/FastAPI, MongoDB ElasticSearch, Redis, Kafka Tools & Platforms: Airflow, MLflow,, AWS, Docker, Kubernetes Technical Skills: Machine Learning, Data Science, LLMs, Agentic AI, Data structures and Algorithms, Software Engineering What I am interested about: Building solutions using State-of-the-Art models is exciting and fun. understanding my comprehension of the core principles underlying ML and data science. Taking open lectures((MIT, Stanford, etc.) on Mathematics and DL to dive deep into the HOW and WHY behind these models. You can reach out to me @ naman247p@gmail.com

Experience

4 yrs

Total Experience

1 yr 4 mos

Average Tenure

1 yr 3 mos

Current Experience

Netomi

Data Scientist

Mar 2025 – Present · 1 yr 3 mos · India · Remote

Helping build robust ML systems using RAG, LLMs and Agentic Framework
Built an Agentic Eval Framework combining BFCL_EVAL3 + LLM-based eval. Standardized the benchmarking process and cut evaluation time for new model releases by 30–40%.
Improved interpretability by 25%, which helped teams focus efforts and boost overall agentic system accuracy by 5%.
Enhanced the RAG (Retrieval-Augmented Generation) system’s accuracy by 5% by implementing efficient document chunking using Markdown-based and Recursive Text Splitters before indexing in Elasticsearch.

RAGLLMsAgentic FrameworkMachine LearningData Science

Lenskart.com

2 roles

Data Scientist

Sep 2023 – Mar 2025 · 1 yr 6 mos · Gurugram, Haryana, India

Frame Removal for Seamless Virtual Try-Ons:
Developed the core ML pipeline enabling realistic virtual eyewear try-ons by removing existing glasses from user selfies using computer vision techniques.
Used GroundingDino and SAM to accurately detect and segment eyeglass regions in user selfies, then applied Stable Diffusion to seamlessly inpaint those areas, restoring natural-looking skin and eyebrow textures.
Improved 20% increase in AR feature adoption and a 5% boost in overall conversions.
Model Photo Generation for different themes using GenAI:
Developed a tool to generate marketing images by enhancing model images with clothes, backgrounds, and poses while preserving the main product.
Implemented product segmentation with GroundingDino and SAM models, and created masks for selective image generation, including inpainting and placement tasks.
Using various text-to-image models like (Stable Diffusion, Flux, Fooocus) to generate different images.
Recommendation System for style and occasion-specific product discovery:
Used LLMs to extract features and evaluate product relevancy for different styles or occasions (for eg; sunglasses for a beach party )
Build a recommendation engine that uses collaborative filtering, LTR, and user attributes (age and gender) to personalize the results for each user.
EDD (Estimated Date of Delivery) Model:
Built an ML solution using XGBoost to provide more accurate delivery date predictions, aiming to reduce customer support calls and improve conversions; developed a backend service with Python and Flask to deploy the model and expose it via API for seamless integration.
Reduce the cost incurred in call queries regarding delivery by 30% and 2+ days under commitment by 20%.
Personalized Couponing System:
Built an ML framework to predict customer purchase intent, correctly identifying 30% of users likely to buy within 7 days; used these predictions for targeted couponing, leading to increased revenue.

Computer VisionGroundingDinoStable DiffusionLLMsXGBoostFlask+2

Data Science Intern

Sep 2022 – Sep 2023 · 1 yr · Gurugram, Haryana, India

Color Swatch Generation using ML
Developed an ML model utilizing computer vision techniques to automate color swatch generation by detecting the colors present in eyeglass frame images.
Achieved 95% accuracy in detecting colors and generating color swatches which led to a 10% increase in CTR (Click-through rate) by reducing the number of clicks for customers to find desired product.
Successfully eliminated all the cost and efforts associated with the manual process of color swatch generation.
Incorporated scripts using Python and Bash to upload generated results into the database.

Computer VisionPythonBashMachine Learning

Thirstysea ltd

Web Development Intern

May 2022 – Aug 2022 · 3 mos · Remote · Remote

Responsible for writing code using KaTeX (Javascript library to display mathematical notations) that enabled server-side rendering and increased the website load time by 5%.
Collaborated with team members using version control systems such as Git to organize modifications and assign tasks.
Used Gatsby, an open-source static site generator, to build the site locally to review the changes made.

JavaScriptGitGatsby