SuyashVakhariya
I build production AI systems and ship products. From ML pipelines to user-facing applications — bridging research and real-world impact.
About
I'm Suyash Vakhariya, an AI Engineer and Technical Product Manager who builds at the intersection of machine learning and product design.
My work spans from neuromorphic computing research and deep learning systems to shipping full-stack web applications that real users interact with daily. I've built 6+ AI/ML systems — including spam detectors, fake news classifiers, recommendation engines, and network intrusion detection — and deployed 7 production web apps on Vercel.
I bring a product-minded approach to every technical challenge: understanding user needs, writing clear specifications, and shipping reliable systems that work at scale. Whether it's training a spiking neural network or designing a cloud security dashboard, I obsess over both the engineering rigor and the user experience.
AI/ML Engineering
Neural networks, NLP, computer vision, and spiking neural network research.
Product Development
Shipping full-stack applications from idea to production on Vercel.
Full-Stack Apps
React, Next.js, Node.js, Python — end-to-end development across the stack.
Technical Leadership
Leading projects, writing PRDs, and collaborating across research and engineering.
Projects
From AI research to production web apps — built, shipped, and deployed.
Featured — AI / ML / Systems
Neuromorphic Computing SNN
Bio-inspired spiking neural network architecture mimicking neuronal firing patterns for ultra-efficient intelligent processing and signal classification.
Email Spam Detection System
Machine learning model trained to classify emails and block spam using NLP techniques and probabilistic filtering with high accuracy.
RFID Car Ignition System
NFC/RFID-based smart ignition system allowing vehicles to start using an encrypted identity card with secure multi-factor authentication.
Deployed — Live on Vercel
WorkSphere OS
Enterprise workflow & team management platform with task tracking, internal messaging, knowledge base, scheduling, and role-based user management — powered by JWT auth and MongoDB.
AgentFence
The local security gate for AI coding agents. Detects secret leaks (tokens, API keys), destructive commands (rm -rf, force-push), and unsafe network egress before execution with native Model Context Protocol (MCP) server support.
CloudSecure
Enterprise cloud security platform with real-time threat monitoring, compliance dashboards, and automated incident response workflows.
PricePulse
Smart electronics price comparison across Amazon, Flipkart & Croma. Search, filter, sort, and find the best deals in India.
Smart Study Planner
All-in-one smart study workspace with Pomodoro focus timer, AI revision schedule generator, active recall flashcard vault, and GPA tracker.
EduNet
Interactive engineering & AI learning ecosystem with 8 curriculum tracks, real-time algorithm visualizer sandbox, technical skill quizzes, and resource vault.
More AI/ML — Research & Security
Fake News Detection
NLP-powered misinformation classifier using advanced text analysis and machine learning to identify fake news articles with high precision.
Movie Recommendation System
Collaborative filtering recommendation engine that analyzes user preferences and movie metadata to deliver personalized suggestions.
Network Intrusion Detection
ML-based cybersecurity system that monitors network traffic patterns and detects malicious intrusion attempts in real-time.
AI API Security Monitor
Intelligent API threat detection system that uses ML to identify anomalous API access patterns and potential security breaches.
Research
Peer-reviewed contributions to AI and machine learning research.
FAMM: Future-Aware Adaptive Memory Management Framework for Long-Term Autonomous LLM Agents
This paper presents FAMM, a novel memory management framework designed for long-term autonomous Large Language Model (LLM) agents. Existing LLM-based agents often struggle with context limitations, memory degradation, and inefficient retrieval during prolonged interactions. FAMM addresses these challenges through adaptive memory organization, future-aware context prioritization, intelligent retrieval mechanisms, and dynamic memory optimization to improve long-term reasoning and task continuity.
Adaptive Memory
Dynamic memory organization that prioritizes relevant context across extended agent interactions.
Future-Aware Prioritization
Predictive context selection that anticipates future task needs for efficient reasoning.
Intelligent Retrieval
Smart retrieval mechanisms that reduce memory degradation in long-running LLM sessions.
Scalable Architecture
Framework designed for real-world autonomous AI agents operating in extended environments.
Skills & Technologies
AI & Machine Learning
Full-Stack Development
Cloud & Infrastructure
Embedded & IoT
Product & Data
Experience & Impact
Neuromorphic Computing Research
Developed spiking neural network architectures for bio-inspired computation — bridging neuroscience and machine learning for energy-efficient AI.
6+ AI/ML Systems Built
Shipped production ML models: spam detection, fake news classification, movie recommendation, network intrusion detection, and API security monitoring.
8+ Production Apps & Tools Deployed
Designed and deployed full-stack web applications and developer tools — including AgentFence (MCP security guardrail), WorkSphere OS, CloudSecure, and PricePulse.
AI-Powered Security & Guardrails
Engineered AgentFence (MCP security guardrail for AI coding agents), network intrusion detection, and API security monitors to intercept leaks and malicious actions.
IoT & Embedded Systems
Engineered RFID-based car ignition system with encrypted identity authentication — bridging hardware (C++/Arduino) and software security.
Full-Stack Proficiency
End-to-end development across React, Next.js, Node.js, Python, TypeScript, and C++ — from ML backends to interactive 3D frontends.
What I'm Working On
Get in Touch
Open to AI engineering roles, technical product management, and research collaborations.