Geetansh Saxena

At PayU's Automation team, took end-to-end technical ownership of an enterprise AI analytics initiative, spanning system architecture, multi-agent orchestration, retrieval engineering, data security, analytical workflows, and cloud deployment. Key contributions: • AI Systems Architecture & Technical Ownership: Architected a modular, 9-node multi-agent system using LangGraph, LangChain, and Toqan LLMs. Defined agent responsibilities, orchestration patterns, state transitions, and execution boundaries to s

At Bosch Research, worked across Machine Learning Engineering, AI Research, and Software Engineering. Took ownership of ML experimentation, model evaluation, developer tooling, CI/CD, and workflow automation. Machine Learning Engineering & Research • ML Systems Architecture & MLOps: Designed modular training and evaluation pipelines using Python, ROS 2, Gymnasium, and Stable-Baselines3. Integrated Optuna for hyperparameter optimization and TensorBoard for experiment tracking. • Model Training & Benchmarki

At Google, worked as a Student Research Assistant with Dr. Jeremy Miles on statistical modeling, simulation-based research, and evaluation methods for large-scale data annotation. Took end-to-end ownership of research design, implementation, experimentation, statistical analysis, and technical documentation. · Statistical Modeling & Simulation: Designed and developed a configurable simulation framework in R to generate synthetic datasets. Modeled real-world data variability, class imbalance, sampling bias,

· Stack: Python (Django), Google Cloud/Azure, Docker, MySQL · Co-led the design and development (backend) of Luna; an academic insights platform, focusing on predicting student dropout · Developed pototype led to funding by Ministry of Science and Arts; poised for adoption by universities across Germany · Oversaw end-to-end deployment of microservices, using Docker and VMs; maintaining best CI/CD and DevOps practices

· Built and Deployed 20+ REST and gRPC APIs. Stack: Go (Gin, gRPC-Go, GORM, SQLBoiler), AWS, Docker, MySQL · Developed a CI/CD Pipeline combining Github Actions, Docker, AWS ECR+EC2 for automated deployment on a push tag · Worked on the development of a Sanskaar Token Reward System for community interactions on Twitter and Discord

· Worked with the Generative Media team at Totality · Used Reinforcement Learning & GAIL (via MLAgents) to develop intelligent NPCs for the Zion Metaverse · Developed Computer Vision models to generate personalized skins, dynamic NFTs, and other digital assets

· Responsible for mentoring and management of Machine learning, Deep learning & Data science projects across various teams · Guided multiple interns across projects; provided technical support and helped perform, present & publish their experiments

· Read multiple papers in the color science domain; implemented proposed color models; added them to the Colour package · Implemented parsers for UPRTek and Sekonic pseudo-XLS and -CSV files; wrote tests; increased code coverage by 12%

· Researched a novel method to estimate utility-pole height and distance using a single image · Implemented various Deep Learning and Automation pipelines; performed wrangling and munging of collected data

· Worked on developing techniques for the application of Deep Learning in the cryptanalysis of lightweight block ciphers · Developed techniques to achieve a significant reduction in required sample size to 2^14 compared to 2^17 in classical distinguishers

· Founded a not-for-profit research start-up with the goal to develop a wearable device that supports the visually impaired in areas like virtual cognizance, navigation support, ease & safety, using the latest developments in AI and Computer Vision · Developing both hardware and software parts of the solution in-house, encompassing features such as Voice Assistant, Object Detection, Image Captioning, Voice-Assisted Navigation, Signboard Detection, Pothole Detection, Face Recognition, etc.

· Revamped 3 expert-level certification courses: Certified Bitcoin Expert, Ethereum Developer, and Hyperledger Developer · Refined content, developed new projects in blockchain and cryptography, tuned to the latest developments in the field

· Worked on the development of a blockchain-based insurance claim processing application using Hyperledger · Actively involved in brainstorming, planning, and development of projects

Class Representative All India Rank 21 in Entrance Examination