AI Product Features
From LLM prototypes to reliable production endpoints.
AI-ML / Software Development EngineerOpen to projects and consultations
Building intelligentand Scalablesystems.
I help teams ship products with modern AI-ML and scalable production-grade software architecture.
Core build tracks I typically deliver for product teams:
Software Core
High-throughput APIs and resilient service architecture.
MLOps + Reliability
Model lifecycle and operational safety from day one.
AI-ML engineering and software execution across data, models, APIs, and production infrastructure.
What I bring
I work with founders and product teams as an AI-ML engineer and software engineer, turning ideas into reliable production systems with clean architecture and strong observability.
I focus on practical delivery, clear interfaces, and systems that teams can ship, debug, and extend without unnecessary complexity.
- AI and machine learning systems from prototype to deployment
- Scalable APIs with robust data contracts and integrations
- Monitoring, alerting, and cost-aware optimization for production workloads
Current Focus
Shipping LLM features, retrieval systems, and event-driven software services for teams that need practical machine learning engineering and production reliability.
Open to freelance projects, consulting, and software engineering roles where strong execution matters.
- RAG quality tuning, grounding, and citation-aware responses
- Low-latency inference with caching and queue-based software execution
- Production observability for model quality, drift, and API health
Technologies I use to build AI features and resilient software systems.
AI and ML
Model development, serving, and evaluation
PyTorchTensorFlowScikit-learnKerasOpenCVLangChainLangGraphCrewAIOpenAIGoogle ADKVertexAI
Backend, API and Data
Service APIs and data-intensive workflows
PythonTypeScriptJavaScriptFastAPIDjangoFlaskExpress.jsPostgreSQLMongoDBRedisMySQLFirebaseBullMQ
Frontend and Product UI
Interfaces for AI products, dashboards, and production tools
TypeScriptJavaScriptReact.jsNext.jsTailwind CSSViteTkinterKivy
Infrastructure and Ops
Deployment, orchestration, and observability
DockerKubernetesCircleCIGitHub ActionsAmazon Web ServicesGoogle Cloud PlatformMicrosoft AzureLangFuseLangSmith
Recent systems built for performance, reliability, and measurable impact.
Web Application
Research-AI
Developed a full stack Deep Research platform for generating citation-grounded long format research documents with multi-agent architecture.
Bench55.32DeepResearch score
ModeMulti-agentSearch + synthesis graph
OutputCited docsGrounded long-form reports
Benchmarked on DeepResearch Bench (Feb 2026) with an overall score of 55.32 beating OpenAI, Gemini and Perplexity DeepResearch.
PythonFastAPIReact.jsLangChainFirebaseOAuth
Desktop Application
Computer Use Agent
Multimodal agent that can perform tasks on a computer based on natural language instructions using vision and tools.
RuntimeLocalOffline-capable mode
Vision2 backendsGUI-Actor + OmniParser
ControlMouse/keyPerception-to-action loop
Can run 100% locally or use a hybrid approach with local and cloud-based LLMs to perform long and complex tasks
PythonMultimodalLocal LLMsOpenCVLangChain
Internal Platform
Internal Multi-Agent Platform
Built and deployed a microservice-based multi-agent platform with real-time RAG pipeline, multiple data source integrations and sessioned chats.
Latency<7sEnd-to-end agent path
AccessRBACGoogle SSO dashboard
DataRAGEnterprise connectors
E2E latency of <7s; custom analytics and monitoring dashboard; RBAC and SSO integration for the dashboard and agents
Comoto Holdings, USA
GCPLLMOpsPostgreSQLFastAPILangChainReact.js
Web Application
Kalvium App Suite
Maintained and enhanced multiple production applications serving 3,000+ concurrent users and services handling 100+ requests/sec.
Users3,000+Concurrent app load
Traffic100+Requests per second
QueueBullMQEval job reliability
Built real-time browser enrollment tracking with Redis fingerprint validation and improved assessment eval reliability across GCP Cloud Run jobs and BullMQ pipelines.
Kalvium
GCPPostgreSQLRedisBullMQReact.jsNext.jsPythonLangFuse
Model Deployment
Health Device Information Classification
Trained and evaluated 120+ classification models on sensor/health signals; compared against baselines and iterated on features and validation.
Models120+Classification runs
Accuracy~85%Average across devices
Records10k+Processed readings
Average accuracy of ~85% across multiple device types; processed 10,000+ records; deployed models with scheduled retraining pipeline.
Samsung R&D Institute India - Bengaluru
PythonMulticlass ClassificationData ProcessingMLOps
Freelance AI-ML engineering and software engineering support for production teams.
AI/ML Engineering
LLM features, RAG systems, prompt pipelines, evaluation frameworks, and production-ready AI workflows.
Software Architecture
Scalable API design, domain modeling, auth patterns, service boundaries, and maintainable software systems.
MLOps and Model Serving
Model versioning, deployment automation, inference optimization, and reliable serving for production ML systems.
Data Pipelines and Integrations
Batch and streaming pipelines, ETL orchestration, and system integrations that keep production data moving.
Observability and Monitoring
Telemetry, alerts, dashboards, and SLO tracking across APIs, AI services, and ML workloads.
Common questions before we build.
What kinds of clients do you usually work with?
Mostly SaaS teams, AI startups, and product companies that need production-grade AI, software, and ML execution without expanding permanent headcount.
Can you work with our existing API stack?
Yes. I usually integrate with existing services and improve architecture incrementally, unless a clean rebuild is clearly the better long-term option.
Do you also handle ML deployment and monitoring?
Yes. I cover model serving, deployment workflows, metrics, logging, and alerting so ML features stay reliable after release.
How quickly can we start?
Usually within 3 to 7 days depending on scope and timeline. Architecture audits can often begin within 48 hours.
Contact
Need help shipping an ML feature or scaling your software platform?
Share your use case and current stack, and I'll send a focused technical plan within 24 hours.
Or email me at:
nabhpatodi1005@gmail.com
Schedule A Call