Manan Patel
Software Engineer - Applied AI
Passionate about building large-scale distributed systems and leveraging cutting-edge Applied AI Engineering to solve complex problems. Currently building agentic systems and real-time ML inference at Amotions AI, with expertise in production-grade AI applications.
About Me
Hi, I'm Manan!
I'm a Software Engineer - Applied AI currently working at Amotions AI in San Francisco (Remote), building large-scale distributed systems, scalable smart systems, and infrastructure for real-time ML inference and agentic systems. With a Master's in Computer Science from California State University, East Bay (GPA: 3.57) and a Bachelor's in Computer Engineering from Gujarat Technological University (GPA: 3.8), I specialize in software engineering, ML engineering, distributed systems, and scalable infrastructure.
Currently, I'm developing agentic systems and real-time ML inference pipelines using Python, C++, Spark, and SQL. My work involves building MCP server-based multi-agent frameworks, creating FastAPI microservices with Redis and AWS ECS, and implementing large-scale distributed systems that improve inter-agent latency by 35%. Previously at Paul Mason Consulting, I implemented Python inference services that made checkout processes 45% faster, designed fraud detection systems with Isolation Forest, and delivered ML microservices that boosted cart conversions by 25%.
I'm passionate about leveraging Applied AI Engineering, Agentic Systems, and Real-Time ML Inference to solve complex problems. My expertise spans from building scalable RAG systems to designing distributed microservices with Redis caching, all while ensuring high performance and reliability in production environments.
Technical Skills
Programming Languages
Databases
Backend Development
Frontend Development
AI / ML & Gen AI
Decision Systems & Analytics
Infrastructure & DevOps
Cloud Platforms
CODING
LeetCode Progress
Tracking my problem-solving journey and algorithmic skills development.
INTEGRATIONS
Seamless Integration For Smarter Workflows
Your agents, tools, and data sources connected through a calm, resilient backbone — from Slack and Drive to AWS, Notion, and more.
Tech Radar
Interactive visualization of technical skills proficiency. Each point represents a technology, positioned by skill level and category.
ML & Gen AI Engineering
35%
Inter-Agent Latency Reduction
Real-time ML inference optimization
25%
Cart Conversion Boost
ML-driven personalization
45%
Checkout Process Speedup
Parallel ML inference services
85%
Medical Bill OCR Accuracy
Computer vision model
Generative AI & Agentic Systems
Multi-Agent Systems
- MCP server-based agent coordination
- Observer, Planner, Executor agents
- Context-aware agent collaboration
- LangChain agent orchestration
RAG Systems
- Document retrieval with HNSWlib
- Vector similarity search
- LLM-powered Q&A systems
- Embedding management & optimization
Gen AI Applications
- GPT-4 powered workflows
- Anime GPT-2 storyteller
- Real-time emotion detection
- LLM-based meeting insights
ML Engineering Pipeline
Data Collection
Data Collection
Gathering raw data from multiple sources including databases, APIs, and real-time streams. Ensuring data quality, completeness, and proper schema validation.
Preprocessing
Preprocessing
Cleaning, normalizing, and transforming raw data into a format suitable for model training. Includes feature engineering, handling missing values, and data augmentation.
Model Training
Model Training
Training machine learning models using frameworks like PyTorch and TensorFlow. Optimizing hyperparameters, implementing early stopping, and managing experiment tracking with MLflow.
Evaluation
Evaluation
Assessing model performance using metrics like accuracy, precision, recall, and F1-score. Conducting cross-validation, A/B testing, and analyzing model behavior on test datasets.
Deployment
Deployment
Containerizing models with Docker, orchestrating with Kubernetes, and deploying to cloud platforms like AWS ECS. Implementing CI/CD pipelines for automated, reliable releases.
Monitoring
Monitoring
Tracking model performance in production using Prometheus and Grafana. Monitoring latency, throughput, prediction drift, and system health. Setting up alerts for anomalies.
Tools: MLflow, Prometheus, Grafana, OpenTelemetry, Pytest, Docker, Kubernetes, CI/CD
Featured Projects
Building production-grade ML systems, Gen AI applications, and agentic architectures. 13+ projects showcasing expertise in real-time inference, multi-agent systems, and RAG implementations.
FlowAgent - Multi-Agent Productivity Platform
ProductionBuilt FlowAgent, a multi-agent productivity system that automates scheduling and follow-ups using LangChain, MCP, and GPT-4. Created Observer, Planner, and Executor agents collaborating via an MCP server for adaptive workflow automation. Developed FastAPI microservices with Redis memory and a React dashboard, deployed on Render.
Document Search Assistant using RAG
Large-ScaleBuilt a large-scale retrieval system optimized for relevance, latency, and cost efficiency. Implemented HNSWlib vector search with LangChain and Hugging Face for automated document retrieval and intelligent question-answering. Evaluated ranking quality and downstream decision impact using retrieval and response-level metrics, achieving sub-100ms query times on 50,000+ documents.
Rental Management Service
Designed a responsive frontend with Next.js (ReactJS), improving page rendering time by 50ms. Built backend APIs in Python FastAPI with MongoDB, enabling location-based search, calendar, and group-size filters to enhance user experience. Implemented authentication and hosting workflows enabling users to manage property listings.
LodgingHub: Full-Stack Platform
Engineered a full-stack web application replicating Airbnb's core features with React, Node.js, and Express. Designed RESTful APIs with payment gateway integration and optimized MongoDB for scalable data storage.
Vector Similarity Search Engine
Built a high-performance vector similarity search engine using HNSWlib and FastAPI, achieving sub-millisecond query times on 10,000+ documents. Deployed with AWS S3, Lambda, and Kubernetes for horizontal scalability.
MedScan: Medical Bill Interpreter
Trained an object detection model on 6,000+ medical bills using TensorFlow and OpenCV, achieving 85% accuracy in text extraction. Created a mobile app with Android Studio for seamless user interaction.
SmartSplit: Expense Tracking App
Built a Flutter-based expense tracking app with Firebase, featuring custom splits, OCR receipt scanning, real-time sync, and push notifications. Implemented smooth UI animations for intuitive UX.
Founder Re-Engagement Decision System
Decision IntelligenceBuilt a decision intelligence and ranking system that converts unstructured communications into structured economic signals (intent, recency, expected response probability). Modeled timing, frequency, and prioritization trade-offs to maximize long-term engagement while minimizing outreach fatigue. Enabled automated experimentation by logging decisions and outcomes to support counterfactual and causal analysis.
AegisChain: Supply Chain Security AI
AI Agents for Supply Chain Code Security - Cloud Run Hackathon Project. Developed intelligent agents to monitor and secure supply chain code, detecting vulnerabilities and ensuring code integrity across the supply chain.
Flask ML Services
Containerized Flask ML service with Docker, MLflow, and automated CI/CD pipeline. Built production-ready ML microservices with model versioning, experiment tracking, and seamless deployment workflows.
Stock Price Prediction
Predicting stock prices using LSTMs and dynamic RNNs. Built deep learning models for time series forecasting with recurrent neural networks, achieving accurate predictions for financial market analysis.
Crop Disease Detection
Computer vision system for detecting crop diseases using deep learning. Built image classification models to help farmers identify plant diseases early, improving crop yield and agricultural productivity.
Anime GPT-2 Storyteller
Creative AI storytelling system using GPT-2 fine-tuned on anime narratives. Generates engaging stories and narratives in the anime style, demonstrating advanced NLP and text generation capabilities.
Experience & Education
Software Engineer - Applied AI
Amotions AI - San Francisco, California, United States · Remote
Software Engineering • ML Engineering • Infrastructure • Distributed Systems • Scalable Smart Systems
Technologies: Python, C++, Spark, SQL, REST APIs, Linux, Pytest
- Built a real-time marketplace decisioning system that optimized content allocation under latency, diversity, and fatigue constraints, analogous to ads delivery optimization
- Designed utility-based ranking and allocation logic balancing short-term engagement with long-term retention and saturation effects, explicitly modeling relevance–exposure trade-offs
- Implemented auction-like prioritization mechanisms to resolve competing candidates under limited surface area, incorporating pacing, exposure caps, and budget-style constraints
- Productionized the decisioning system and instrumentation to support online A/B testing, counterfactual analysis, and iterative policy tuning in a low-latency production environment
- Building AI systems for real-time emotion and speaker detection to enhance LLM-based meeting insights using OpenAI, Gemini, and Pinecone
Master of Science in Computer Science
California State University, East Bay - Hayward, CA
GPA: 3.57
Coursework in Operating Systems, Distributed Systems, Machine Learning, Computer Vision, and Natural Language Processing. Focus on advanced distributed systems, AI/ML algorithms, and scalable system design.
Software Engineer
Paul Mason Consulting - Vadodara, India
Clients: Crew Clothing, Morrisons
- Built ranking, pricing, and allocation pipelines over large-scale retail transaction and browsing data, improving decision accuracy by 25%
- Designed feature engineering workflows modeling customer response, retention, elasticity, and seasonality to support econometric and ML analysis
- Integrated real-time ML inference into backend services supporting 500K+ daily events, meeting strict latency and availability requirements
- Implemented experiment design, causal analysis, drift detection, and data quality checks, reducing model-related production incidents by 20%
- Delivered a "Frequently Bought Together" recommendation microservice integrated into a monorepo ecosystem, applying ML-driven personalization that boosted cart conversions by 25%
Software Engineer Intern
Paul Mason Consulting - Vadodara, India
- Built a FastAPI service to process real-time user events and persist results in PostgreSQL, exposing REST APIs that enabled other services in the retailer's checkout pipeline to access customer activity data
- Containerized and deployed the service with Docker on AWS EKS, making it scalable and easier for the platform team to manage in production
- Integrated OpenTelemetry monitoring to capture latency and error metrics, giving engineering teams visibility into bottlenecks and helping ensure reliable customer checkout flows
Software Engineer Intern
TenUp Software Services - Vadodara, Gujarat, India · Hybrid
Developed and tested API contracts and wrote integration tests for backend services in a cloud-based project management platform, while learning CI/CD and Docker for reliable feature delivery.
Technologies: Python, Production Deployment, API Testing, CI/CD, Docker, Integration Testing
Bachelor of Engineering in Computer Engineering
Gujarat Technological University - Ahmedabad, India
GPA: 3.8
Strong foundation in computer science fundamentals, data structures, algorithms, and software engineering principles. Graduated with comprehensive knowledge in distributed systems and scalable architectures.
Get In Touch
Let's build something amazing together!
I'm always interested in discussing new opportunities, collaborating on innovative projects, or chatting about the latest in distributed systems and AI/ML. Feel free to reach out!