Hitanshu Kumar Singh

Co-Founder

Bengaluru, Karnataka, India0 mo experience

AI EnabledAI ML Practitioner

Key Highlights

Reduced video processing latency by 95%
Achieved sub-200ms response times for cloud cost monitoring
Built autonomous agents for lead generation automation

Stackforce AI infers this person is a Backend & AI Engineer specializing in SaaS and Fintech solutions.

Contact

Skills

Core Skills

Event Driven ProgrammingAi EngineeringBackend Web Development

Other Skills

Retrieval-Augmented Generation (RAG)Asynchronous workFFmpegFastAPIAsyncIOData ValidationStatistical Moving AverageREST APIsScalabilityVector DatabasesLarge Language Models (LLM)DockerLLaMA-3Postman APIGit

About

I build the engines that power applications, not just the interface. I am a Python-first Backend & AI Engineer focused on Event-Driven Architecture and GenAI Automation. 🔧 What I Build: Performance: Engineered a video processing pipeline (Veloce-AI) using FFmpeg and AsyncIO, reducing processing latency from 45s to 2s and optimizing memory usage for low-resource (512MB RAM) environments. Data Logic: Developed a cloud-cost monitoring backend (Kumo) using FastAPI, implementing statistical algorithms (Moving Average) to detect financial anomalies with sub-200ms response times. Automation: Built autonomous research agents (ColdLeads AI) using LangChain and LLMs to automate complex lead generation workflows. 🛠️ Tech Stack: Python, FastAPI, AsyncIO, LangChain, LLMs, PostgreSQL, REST APIs. I solve problems related to Latency, Resource Constraints, and Backend Logic. If you’re looking for robust systems or AI-driven automation that actually works, let's talk.

Experience

0 mo

Total Experience

Average Tenure

Current Experience

Veloce-ai

Independent Builder

Feb 2026 – Mar 2026 · 1 mo · Bengaluru, Karnataka, India · Remote

Engineered a high-concurrency video processing engine, slashing data processing latency by 95% (45s → 2s) through a hybrid ingestion pipeline using FFmpeg.
Architected a memory-efficient system that eliminated OOM crashes in strictly constrained 512MB RAM environments using asynchronous FastAPI Background Tasks.
Developed a scalable RAG (Retrieval-Augmented Generation) pipeline with Pinecone and Llama-3, achieving sub-second inference for real-time natural language video querying.
Enforced strict data integrity via Pydantic schemas, reducing runtime type errors by 40% during concurrent heavy-payload uploads.

Retrieval-Augmented Generation (RAG)Asynchronous workFFmpegFastAPIEvent Driven ProgrammingAI Engineering

Kumo finops

Backend Developer

Jan 2026 – Feb 2026 · 1 mo · Bengaluru · Remote

Designed a high-throughput RESTful backend achieving sub-200ms latency for real-time cloud cost anomaly detection.
Optimized data processing efficiency by 80% by implementing a custom Statistical Moving Average algorithm, outperforming heavy ML models in runtime speed.
Integrated an event-driven notification architecture using webhooks to deliver instant Slack/Teams alerts with zero downtime.
Stabilized database performance by managing SQL sessions via FastAPI Dependency Injection, preventing connection leaks under high concurrency.

FastAPIData ValidationBackend Web DevelopmentAI Engineering