Hi, I'm Vidit.
I'm a Data Scientist
I make impactful AI/ML based solutions that take things to the next level. I also create interesting self projects in my spare time and do problem solving on Leetcode. Let's connect!
About.
Hey there! I'm Vidit Singh Negi, and I am a Data Scientist currently working at Clarivate. I Studied Computer Science(Btech) from Jaypee Institute of Information Technology, Noida.
I specialize in building AI/ML Solutions and Agentic Workflows, with a focus on development, optimization and Deployement. I'm passionate about creating clever and impactful models that provide great value in the industry.
I code in Python and C++, and have experience with Langchain/Langgraph Ecosystem, Pytorch, Tensorflow, Flask, FastAPI and various data processing libraries such as Panda, Numpy, OpenCV and more. I've also worked with services provided by AWS and Azure focusing on AI developement, like SageMaker, Bedrock, Guardrails, Step Functions, Lambda, SQS, and more. I Also have experience with Databases like Aurora Postgres, Redshift, MongoDB, and vector databases like Qdrant, Chroma and Pinecone.
When I'm not coding, I enjoy spending time making music and playing instruments. I believe that maintaining a healthy work-life balance is crucial for staying productive and motivated.
I'm always looking for new challenges and opportunities to learn and grow as a developer. If you're interested in working together or have an opportunity that might be a good fit for me, please feel free to reach out! 🔗
Used at work
Learning
Projects.
Experience.
Build an end-to-end Text-to-SQL based Agentic Chatbot for Derwent Patent Analytics. Utilized Langchain/Langgraph, MLflow and AWS Services for developement. Used Promptfoo And Ragas for automated security assessment. Made multiple RAG based tools, secured the chatbot by using AWS bedrock Guardrails, integrated buffered streaming for smooth user experience and deployed on ECS Cluster. Worked with AI Classifier, a massive project on AWS, used Step Functions for orchestration, Sagemaker for BERT hierarchical training and Inference, also used Lambda and SQS to make the pipeline.
Owned many projects and collaborated across multiple teams. Researched, built, trained, optimized and deployed GenAI models including Diffusion models, GANs and Transformers, using Pytorch, FastAPI, docker, Wandb, AWS/Azure. Made Bert-Diffusion Model for font generation, reduced designing time by 30% for designers. - Patent. Built custom 3-staged diffusion model for Japanese fonts generating characters with >90% IOU (Improved by 15%).
Worked with building POS systems. Built REST-APIs with Flask, utilized MongoDB for data storage, did automation using Selenium, and trained ML models for classification. The automation pipeline saved many hours of manual sales report extraction, improving productivity by 25%. Made Flask APIs to handle requests, worked with MongoDB and AWS S3 to store the sales reports.
Worked on ANPR system and made Abandoned Object Detection System with object detection, OCR, YOLOv8 and YOLO-NAS, DeepSort, Background Subtraction, Depth Estimation, Implemented Kafka. Optimized Model inference speed using TensorRT and ONNX and made it 4x faster. Built Docker containers to deploy on toll plaza sites. Increased the ANPR accuracy from 85% to 93%.
Contact.
Send me an email if you want to connect! You can also find me on Linkedin.