AI Engineer with 1+ years in Computer Vision & Deep Learning pipelines
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
ML researcher and engineer with experience building deep learning pipelines across computer vision and time-series domains. Hands-on with PyTorch, transformer attention mechanisms, and model interpretability with published work on CNN architectures and attention-augmented detection models. Interested in XAI, mechanistic interpretability, and making model decisions legible.
Bennett University
B.Tech. · CSE
August 1, 2022 – June 30, 2026
APBIE Board
Senior Secondary
June 1, 2020 – May 31, 2022
CBSE
Secondary
N/A – May 31, 2020
Transvolt Mobility
Data Science Intern
January 1, 2025 – Present
India
Critic AI
June 1, 2026 – Present
Built a production-style RAG pipeline with FastAPI serving contextual game review summaries – document chunking, embedding generation, and vector similarity retrieval over ChromaDB. Integrated OpenAI API with structured prompt engineering to produce sentiment-aware outputs (pros/cons, verdict) from retrieved context mirroring the LLM provider integration pattern used in real agent backends. Designed the retrieval layer to be provider-agnostic swapping embedding models or vector stores required changing a single config, not rewriting pipeline logic.
F1 Brand Intelligence
June 1, 2026 – Present
Shipped an end-to-end logo detection and brand visibility scoring pipeline from raw video ingestion through frame extraction, object detection, temporal tracking, and an interactive analytics dashboard. Designed a novel Pixel-Second Score (PSS) metric weighting logo visibility by frame centrality, detection confidence, and attention habituation a custom scoring engine layered over raw inference output. Implemented a ByteTrack tracker with Kalman filtering to bridge detection gaps caused by occlusion, and an active learning ranker (uncertainty + diversity + class coverage) to cut annotation effort by ~60%.
AgriFind
June 1, 2026 – Present
Designed and deployed CNN model with 95% accuracy on 40,000+ agricultural leaf images for real-time crop disease detection. Implemented data augmentation and preprocessing to improve model reliability. CNNs are ideal for image classification tasks, and Streamlit offers a fast deployment route for machine learning applications without needing a separate frontend. Deployed via Streamlit, enabling farmers to access low-cost AI-powered diagnosis tools
ECG-Image Based Heartbeat Classification for Arrhythmia Detection
June 1, 2026 – Present
Proposed a CNN-based deep learning model for arrhythmia classification using 2D ECG images, achieving 98% accuracy across multiple arrhythmia classes. Developed an end-to-end pipeline integrating image conversion, noise reduction, and real-time arrhythmia detection from raw ECG signal to classified output.
Enhancing YOLO-Based Detection for Small Objects in Aerial Imagery Using Multi-Scale Attention and Dynamic Heads
January 1, 2025 – Present
Proposed targeted enhancements to YOLOv8 integrating multi-scale attention mechanisms and dynamic detection heads to address small object detection failures in aerial and drone imagery. Conducted ablation studies on attention head contributions across detection scales analyzing which spatial regions and feature channels drove small-object recall improvements, surfacing interpretable failure modes in baseline YOLOv8.
Algorithmic Toolbox
UC San Diego
June 1, 2026 – Present
Convolutional Neural Networks
DeepLearning.AI
June 1, 2026 – Present
Neural Networks and Deep Learning
DeepLearning.AI
June 1, 2026 – Present
Cultural Fit Analysis
The candidate's diverse project portfolio, spanning RAG pipelines, computer vision, and time-series analysis, indicates a broad interest in AI applications. Their academic background and certifications show a commitment to continuous learning and staying updated with cutting-edge AI research. The personal projects demonstrate initiative and a passion for applying AI to solve practical problems, which aligns with an innovative and growth-oriented culture.
Soft Skills & Operational Fit
The candidate demonstrates strong problem-solving skills through the design of novel metrics (PSS) and active learning strategies. Their project descriptions indicate an ability to work independently on complex technical challenges and a proactive approach to learning new technologies. The focus on provider-agnostic design and real-world deployment suggests an understanding of operational considerations.