AI & Data Science Engineer
Transforming data into intelligent solutions
Passionate AI engineer with expertise in machine learning, computer vision, and natural language processing. Currently pursuing MSc in Data Science while working on cutting-edge AI projects.
Professional Experience
AI Engineer
Antematter • Jun. 2024 – Present
- Developed a code review agent using OpenAI's API and LangChain's ReAct prompting
- Integrated Slither and ItyFuzz for smart contract analysis
- Collaborated on Web3 tool for on-chain data and market stats using mixture-of-agent framework
- Contributed to IAO platform enabling dynamic creation of AI agents
Data Scientist
Spatial Labs Pvt. Ltd. �€¢ Feb. 2023 – Jun. 2024
- Developed hybrid technique combining Learned Noise Prediction and Fourier Transform for image authentication (89% accuracy)
- Improved YOLOv8 model efficacy from 67 mAP to 95 mAP for vehicle monitoring
- Assessed LLM responses using PairRM in LLM Blender
- Implemented Whisper model for offline speech-to-text translation
- Created synthetic data generation scripts in Blender
Data Science Intern
Spatial Labs Pvt. Ltd. • Sep. 2022 – Feb. 2023
- Customized UNET model for land cover classification (IoU 0.92)
- Adapted YOLOv5 for forest tree damage recognition
- Competed in ship detection challenge (mAP 0.86 public leaderboard)
Answers Intern
Educative Inc. • Jun. 2022 – Aug. 2022
- Authored 28 technical pieces on AI and computer vision topics
- Evaluated 52 technical articles
- Covered ViT, Segmentation, Transformers, and Data Augmentation
Featured Projects
TrajectoryQA
System for responding to inquiries based on trajectory descriptions using Geolife dataset and OpenStreetMap. Achieved 79.6% accuracy with GPT-2 model.
Email Lead Generator
Two-agent system using LangChain and GPT-4 for lead generation and follow-up. Opener agent creates initial emails, Escalator handles responses.
Smart Contract Auditor
Code review agent using OpenAI API and LangChain's ReAct prompting. Integrated Slither and ItyFuzz for comprehensive analysis.
Satellite Image Authentication
Hybrid technique combining Learned Noise Prediction and Fourier Transform for distinguishing authentic satellite images (89% accuracy).
