Languages
Python, Java, C++, TypeScript, SQL
Software Engineer | AI & Machine Learning | Robotics
Anuj Wadi is a Software Engineer working across AI, Machine Learning, and Robotics, with an MS in Robotics and Autonomous Systems (AI) from Arizona State University. He builds intelligent systems that combine agentic AI, machine learning, and autonomous systems to solve real-world problems โ from a production AI-assisted Blood Bank Management System deployed across 3 hospitals to a multimodal conversational AI runtime with sub-200ms end-to-end latency. His work spans agentic workflows and RAG pipelines, transformer fine-tuning, and cloud-native backends, bridging research with practical, production-grade deployment across healthcare, finance, and robotics.
Python, Java, C++, TypeScript, SQL
Data Structures & Algorithms, OOP, System Design, Distributed Systems
MCP, Agentic Workflows, RAG, LangChain, LangGraph, Pinecone, Prompt Engineering
FastAPI, Spring Boot, REST APIs, Kafka, MongoDB, Pandas
AWS, GCP, Docker, Kubernetes, Terraform, GitHub Actions, Git
PyTorch, JAX, Transformer Architectures, LoRA/PEFT Fine-Tuning, Pandas, NumPy

Developed a production-ready conversational AI platform with real-time speech I/O and avatar responses, achieving sub-200ms end-to-end latency on FastAPI and LLM APIs. Unified OpenAI and speech APIs behind one async FastAPI backend, supporting concurrent multi-user sessions without degrading response latency. Built with Python, FastAPI, WebSockets, Deepgram, GPT-4o, ElevenLabs, and Tavus.
Led a team of 5 to build a multimodal mental health monitoring platform on LSTM and BERT NLP pipelines, detecting distress signals across 3 data modalities. Implemented privacy-preserving infrastructure with encrypted storage and access controls, reducing per-sample inference time from 1.4s to 0.9s on biometric data. Built with Python, TensorFlow, LSTM, and BERT.

Engineered an autonomous maze navigation system on ROS, OpenCV, and inverse kinematics, hitting a 92% success rate across 50+ trials and trimming collisions from 18% to 4%. Deployed a Gazebo simulation environment that eliminated the hardware dependency, shrinking iteration cycles from 3 hours on hardware to 45 minutes in simulation.

Developed ML-based parking prediction system with full-stack IoT solution, pitched to 3 major investors, achieving 40% reduction in parking search time.

Developed a fully functional website for an AI-powered chalkboard within 3 weeks, leveraging Lovable AI and OpenAI for real-time transcription and summarization to enhance student engagement.

Engineered Q-learning model processing 50K+ daily data points for real-time trading signals, achieving 17% simulated ROI increase over six months.

Built predictive analytics model using Random Forest and Logistic Regression, guiding 129+ students toward institutions with higher admission success rates.

Developed ML model achieving 91% accuracy using Random Forest and SVM with Apache Spark data pipeline for large-scale medical datasets.
Career snapshot
Last updated July 2026
Preview unavailable in this browser.
Use the download button to view the resume.
Feel free to contact me for any work or suggestions. I'm always open to discussing new projects, creative ideas, or opportunities to be part of your vision.