AKSHAY

SANTHOSHKUMAR

Building Agentic AI workflows & Scalable Systems.

About

3 Years of Building in AI and Distributed Systems

AI-Engineer focused on building Autonomous Agentic-workflows and Distributed platforms that scale efficiently while driving down operational Costs.

This Portfolio is about an Engineer who specializes in AI agents and Systems Design. I am always looking for challenges of replacing legacy monolithic systems with event-driven architectures, finding the signal in the noise to leave platforms sturdier, faster, and more automated than I found them.

My professional foundation was built at McAfee India, where I replaced legacy monoliths with event-driven AWS microservices. Scaling systems to process 100K+ events daily and optimizing geographic database sharding turned my curiosity into a deep expertise in performance optimization and production-grade reliability.

At ChannelCore, an early-stage startup, I focused on building AI-driven media kit generators and LLM-powered brand deal workflows. By integrating Next.js with scalable LLM APIs, I automated complex tasks, reducing manual preparation time by over 90% and improving overall efficiency.

Currently, as an AI Engineer at CitiBank in New York, I automate the extraction of complex loan notices using FastAPI, LangGraph, and Python. By designing multi-stage LLM processing pipelines and robust validation architectures, I significantly increase extraction accuracy and reduce manual processing efforts.

Experience

Roles that shaped my Systems thinking from scaling Fault-tolerant Microservices to building autonomous AI-Agents workflows.

CitiBank

AI Engineer

April 2026 - Present · New York, USA

Built a production AI MVP in 6 weeks using FastAPI, LangGraph, and Python to automate extraction of syndicated loan notices through a multi-stage LLM processing pipeline.

Impact: Increased extraction accuracy from ~55% to ~90% by decomposing a monolithic prompt into specialized filtering, event classification, and extraction stages, eliminating cascading prediction errors.

Architecture: Designed a maker-checker validation architecture with structured outputs, independent verification, and retry policies, significantly improving reliability before downstream processing.

Efficiency: Reduced manual processing effort by ~60% by automating event classification and field extraction for complex unstructured loan notices while maintaining deterministic structured outputs for downstream systems.

Skills & Technologies

FastAPILangGraphPythonLLM Pipelines

ChannelCore

AI Engineer | Early-stage startup

Jan - Apr 2026 · California, USA

Built an AI-driven media kit generator using Next.js and LLM APIs to summarize creator analytics and automate brand profile creation, improving workflow efficiency by 30%.

Built LLM-powered brand deal workflows to analyze contracts and generate personalized pitches, reducing preparation time by 92% (from 1 hour to 5 minutes).

Skills & Technologies

Agentic AILLM PipelinesNext.jsRedisSemantic Caching

University at Buffalo - SUNY

Master's in Computer Science

2024 – 2025 · New York, USA

Deepened expertise in scalable systems, machine learning, and advanced algorithm design while balancing research-oriented rigor with practical implementation.

Core: Distributed Systems and Object-Oriented Design

Focus: Machine Learning and Deep Learning

Coursework

Machine LearningDeep LearningDistributed SystemsComputer Vision

McAfee

Full Stack Software Engineer

June 2022 - July 2024 · India

DISTRIBUTED SYSTEMS & PERFORMANCE OPTIMIZATION

Replaced legacy monolithic architecture with an event-driven AWS microservices system, scaling to process 100K+ events using Apache Kafka message streaming and Redis caching infrastructure.

Optimized database performance through comprehensive log analysis and schema refinement, reducing query latency by 75% (from 2s to 0.5s) and lowering CPU operational costs by $2K monthly.

Implemented horizontal scaling through geographic database sharding, supporting 500+ requests per second with 99.9% uptime while achieving $15K annual savings in EC2 infrastructure costs.

TEAM LEADERSHIP & DEVELOPMENT

Mentored 4 software engineers through pair programming sessions and design review processes, reducing team ramp-up time by 40% and decreasing code redundancy by 30% through knowledge sharing initiatives.

Led cross-functional collaboration on microservices architecture design, establishing development standards and best practices for scalable system implementation.

Contributed to engineering culture development through code review processes and technical documentation standards.

Skills & Technologies

MicroservicesKafkaRedisAWSDatabase Sharding

McAfee

Software Developer Intern

2022 · Remote

Strengthened authentication, CI/CD, and frontend observability while improving release velocity and reducing login friction at scale.

JWT/bcrypt auth for 1,000+ concurrent sessions

CI/CD automation reduced release cycle to 5 days

Skills & Technologies

React.jsNode.jsJWTCI/CDAuthentication

Anna University

Bachelor's in Computer Science

2018 – 2022 · India

Built a strong systems foundation through hands-on projects and rigorous coursework in core computer science disciplines.

Core: Data Structures and Operating Systems

Focus: Networking and Cloud Computing

Coursework

Data StructuresObject Oriented ProgrammingOperating SystemsCloud ComputingComputer NetworkingBig Data Analysis

Projects

Autonomously drafts job applications from RSS feeds

Just Apply

Developed an agentic job application system that polls RSS feeds for job links and executes a LangChain workflows per job link. The workflow retrieves the job description, extracts key requirements, and drafts a tailored application using an LLM. This system automates the initial application process, allowing for rapid and personalized job submissions.

1 / 4

Skills

Explore with your mouse
It's interactive

Testimonials

Ratnesh Dubey

Software Architect at McAfee

I had the pleasure of working closely with Akshay when he was assigned to our project at McAfee. During that time, Akshay demonstrated not only exceptional technical expertise but also a remarkable attitude that made him a true asset to the team. As a Node.js developer, his deep understanding of backend technologies, asynchronous programming, and API integrations was evident.

Awards

Leetcode Knight
Ranked within the top 3% of global developers on LeetCode, achieving the prestigious Knight badge through consistent performance in weekly competitive...
Top 3.09% Globally | 1,960+ Problems Solved
Competitive Programming Night
Secured first place in LeetCode Night, a university-wide competitive programming event at the University at Buffalo where participants competed to sol...
1st Place | University at Buffalo Coding Competition
HackWithInfy - Infosys
Qualified for the interview stage by ranking in the top 5% nationwide in a hard-difficulty competitive programming assessment, demonstrating strong ex...
Top 5% | National Coding Challenge
Code Currents '22
Secured third place in a national-level coding competition hosted by the National Institute of Technology, Tiruchirappalli, where participants solved ...
3rd Place | National Competitive Programming

Contact

Perfect

Let's talk

30 min intro call

Let's worktogether

Have a project in mind? I'd love to hear about it. Let's talk.

trevorakshay@gmail.com

3 Years of Building in AI and Distributed Systems