Anirudh Yadav

+•• ••••• •••••••••••@•••••.com

Software Engineer · Intello Labs · 4.1 yrs

As a software engineer with three years of experience, I am driven by a deep ambition to achieve mastery in my craft, approaching complex challenges with a logical and concentrated focus. I thrive on autonomous problem-solving, always seeking to build innovative and high-quality solutions that align with a broader, idealistic vision for technological advancement.

Ask about Anirudh

Skills

Claimed

Dataset ManagementDeep LearningBattery Technology KnowledgeMathematical ModelingMask R-CNN ModelingData PreprocessingTraining Data CreationObject Bounding Box GenerationModel Performance ImprovementML Detection Speed OptimizationMachine Learning Model DeploymentModel Selection Trade-Off Analysis

Proof of Experience

Software Engineer

Intello Labs · Jun 2026 – Present

  • Developed a camera calibration utility for automated fruit sorting systems using ArUco markers, improving camera alignment and focus while reducing manual calibration effort.
  • Developed a Vision-Language Model (VLM) pipeline using Qwen3-VL 32B to automate invoice parsing and information extraction, significantly reducing processing time compared to manual workflows.
  • Developed a FastAPI-based application for loading datasets from Annoflow and training RF-DETR models for defect localization across different agricultural commodities.

Software Engineer

Indian Council of Medical Research · Apr 2025 – Jun 2026

  • Developed and evaluated multiple deep learning models for automated tuberculosis detection using medical imaging data: Custom CNN: Designed a custom architecture achieving 99.57% accuracy.
  • Developed and evaluated multiple deep learning models for automated tuberculosis detection using medical imaging data: Modified AlexNet: Optimized the architecture by removing batch normalization and dropout layers, achieving 95.95% accuracy.
  • Developed and evaluated multiple deep learning models for automated tuberculosis detection using medical imaging data: VGG16: Applied transfer learning and fine-tuning, achieving 99.05% accuracy.
  • Developed a Django-based web portal for real-time TB diagnosis through chest X-ray image upload and automated model inference.
  • Developed an end-to-end object localization pipeline using YOLO to detect mosquito breeding habitats from aerial and ground-level waterbody images for AI-assisted vector-borne disease surveillance.
  • Trained and optimized deep learning models on a confidential in-house dataset of 13,000+ annotated images, incorporating data preprocessing and augmentation techniques to improve robustness across varying environmental conditions.
  • Evaluated model performance using standard object detection metrics including mAP, Precision, Recall and Intersection over Union (IoU), supporting reliable automated identification of potential mosquito habitats.
Jio Platforms Limited

Software Developer

Jio Platforms Limited · Jul 2022 – Jan 2025

  • Led the development of Neural Network models for State of Charge (SoC) and State of Health (SoH) estimation of Lithium-ion batteries for deployment on Edge devices.
  • Designed an end-to-end data processing pipeline using battery parameters including voltage, current, 16 internal temperature sensor values and external temperature for battery state estimation.
  • Performed data preprocessing, feature engineering and experimentation with multiple Neural Network architectures, including MLP and LSTM models, to identify high-performing approaches for battery state prediction.
  • Fine-tuned and evaluated models using experimental and open-source battery datasets, optimizing prediction accuracy and generalization across different operating conditions.
  • Deployed and optimized Neural Network models on Edge devices for real-time inference.

Proof-of-Work

No completed sessions yet. Anything Anirudh finishes shows up here with the evidence behind it.

Proof of Education

B.Tech, Computer Science

Jaypee Institute of Information Technology, Noida · 2018 — 2022

A Proof CV — one profile, kept current, with the proof attached. Back to the portal