Hi, my name is
Maheshwar Kuchana
I build AI Agents
I’m a Senior ML Engineer building production AI systems — from traditional ML to RAGs and multi‑agentic systems, end‑to‑end with strong MLOps practices.
About Me
I'm a Senior Machine Learning Engineer with 5+ years of experience building and deploying production AI systems at scale. I specialise in modern AI frameworks, cloud platforms (Azure, AWS), and MLOps practices across the full product lifecycle.
I've delivered end‑to‑end AI products that optimise business decisions and automate complex processes across healthcare, retail, finance, insurance, and banking — where reliability and measurable impact are critical.
I excel in cross‑functional environments, partnering with product, engineering, and data teams to deliver high‑impact solutions. Recently, I've focused on operationalising multi‑agentic systems with robust evaluation and safety guardrails.
Here are a few technologies I’ve been working with:
- AI Agents
- RAG
- LangGraph
- MCP
- Vector DB
- MLOps
- AWS / Azure
- PyTorch / TensorFlow
Where I've Worked
Senior Machine Learning Engineer @ Faculty AI
December 2024 - Present
- Architected a production‑grade multi‑agent LangGraph platform on Azure Kubernetes Service (AKS) that automates KYC workflows for an international investment bank, cutting manual review time by 70%
- Designed AgentOps toolkit (guardrails, tracing, offline evals, CI/CD via GitLab) enabling weekly releases with one‑click rollback and zero‑downtime migrations
- Implemented an event-driven based custom Agent-to-Agent (A2A) protocol with fault-tolerant orchestration (horizontal autoscaling, circuit-breakers) that coordinates 100+ concurrent agents, sustaining peak load and cutting cloud spend 25%
- Led development of a multi‑agent group‑life claims‑automation system leveraging Azure Document Intelligence and GPT‑powered agents to process 1 000+ claims/day with 99.9% SLA compliance
- Conducted enterprise AI maturity assessments and advised senior leadership on roadmap, driving adoption of secure MLOps best practices across three business units
Some Things I've Built
Featured Project
Machine learning predicts live-birth occurrence before IVF treatment
This work mainly focuses on making predictions of live-birth occurrence when an embryo forms from a couple. Here, we compare various AI algorithms, including both classical ML, deep learning architecture, and an ensemble of algorithms on the publicly available dataset provided by Human Fertilisation and Embryology Authority (HFEA).
- Machine Learning
- IVF
- Deep Learning
- Python
- Ensemble Algorithms
Featured Project
AI aiding in diagnosing, tracking recovery of COVID-19 on Chest CT scans
Coronavirus (COVID-19) has spread throughout the world, causing mayhem from January 2020 to this day. Owing to its rapidly spreading existence and high death count, the WHO has classified it as a pandemic. Model performance is evaluated in terms of F1-Score, Mean intersection over union (Mean IoU). It is noted that the proposed approach results in 97.31% of F1-Score and 84.6% of Mean IoU.
- Covid-19
- Deep Learning
- Computer Vision
- Semantic Segmentation
Featured Project
Fingerprint Matching - An Experimental Approach
An approach has been proposed to find the region of interest (ROI) which is based on dividing the image into blocks. Blocks of 8 images are compared in terms of ‘Histogram of Oriented Gradients (HOG) Descriptor’ and the error is calculated. The group of error points in each block is considered to be a cluster. The cluster which is less scattered would be the region of interest (ROI).
- Fingerprint Analysis
- ML
- Image Processing
Other Noteworthy Projects
view the archiveQuantitative Imaging of the shared placenta in twin pregnancies
We present a quantitative, multi-modality pipeline as a proof-of-concept to map the placenta blood network to a high resolution, and discuss its clinical potential for TTTS Surgery.
Aatmaan
Aatmaan offers Disease Diagnosis, Prognosis using Intelligence Systems and maintain Patient Records that improves, eases the overall clinical workflow and gives immense experience for doctors & patients.
Fooling Neural Networks
Developed ways to fool neural networks to function as per our wish. Modifying weights and biases, Backdooring, Extracting information are the techniques used for this purpose to hack them
Retinal Vessel Segmentation in Fundus Images of Eye
Semantic Segmentation. Developed Convolutional Neural Network which segments blood vessels in Fundus images of human eye retina. Implemented LadderNet.
Online Compiler
Implemented an Online cloud-based compiler which can compile C, C++, Java, Python languages. It can also be used as Platform as a service by deploying in real-time.
Cloud Based Keylogger
A Cloud based Keylogger Services where customer registers his/her email id and add tracker code in victim's PC to get frequent updates of Key Logs on email.
What’s Next?
Get In Touch
I'm actively building production AI systems and multi-agent platforms. Whether you're interested in AI Agents, RAG systems, MLOps, or just want to connect about the latest in ML engineering, my inbox is always open. I'd love to hear from you!
Say Hello