Shamanth

AI Engineer & Frontend — shipping agentic systems, RAG pipelines, and developer UX. I blend product intuition with systems engineering to build fast, reliable LLM apps.

View Workhello@shamanth.dev

Based in San Francisco

Open to remote work

AI Systems + Frontend

RAG, agents, benchmarks

Currently available

Starting mid‑September

Shamanth

Recent Work

View all
RAG dashboard
RAG Platform

Vector‑backed Retrieval

Hybrid search, chunking, schema‑aware re‑ranking, observability.

Agentic workflow
Agents

Agent Orchestrator

Multi‑tool planning, retries, guardrails, tracing via OpenTelemetry.

Code copilot
Developer UX

Code Copilot

Inline suggestions, context windows, evals, and latency budgets.

Data pipeline
Data Pipeline

Real‑time Analytics

Streaming data processing with Apache Kafka and real‑time dashboards.

ML model
Machine Learning

Custom Model Training

Fine‑tuned transformers for domain‑specific tasks with custom datasets.

API system
API Architecture

Scalable Backend

Microservices architecture with GraphQL, Redis caching, and auto‑scaling.

Trusted by product teams

From seed‑stage startups to enterprise platform groups.

TechFlow
Nexus Labs
DataSync
VisionCorp
CloudBase
InnovateTech
FlowState

Tech Stack

Frontend & UI

TypeScriptNext.js 14React 18Tailwind CSSFramer Motion

AI & Backend

Python 3.11+FastAPILangChainOpenAI APIClaude APIvLLMOllama

Focus areas: RAG optimization, agentic workflows, prompt engineering, model evaluation, and production-ready AI systems with sub-second latency.

RAG Pipeline

~42 lines
from fastapi import FastAPI
from rag import embed, search, rerank, answer
from tracers import trace

app = FastAPI()

@app.post("/ask")
@trace("ask")
def ask(q: str, user_id: str):
    q_vec = embed(q)
    chunks = search(q_vec, k=20, filters={"user": user_id})
    ranked = rerank(q, chunks)[:6]
    return answer(q, ranked, tools=["browser", "code"], guardrails=True)

Timeline

  1. 2025

    Independent — AI Engineer

    Building production AI systems, RAG pipelines, and agentic workflows for startups and enterprise teams.

  2. 2022 — 2024

    Senior Product Designer — Analytics

    Led design for data visualization platform, shipped ML-powered insights dashboard used by 10k+ analysts.

  3. 2017 — 2021

    Frontend Engineer — Commerce

    Built responsive e-commerce platform with React/Node.js, optimized for mobile conversion and performance.

Model Evals

Continuous evaluation of prompts, tools, and retrieval quality across production workloads.

Pass@1 by model (weekly)

Live data
Current best
82%
+4% this week
Context win
+10% RAG
vs baseline
Latency p95
780ms
Within SLA

Eval Categories

Code generation87%
Q&A retrieval82%
Tool usage76%
Reasoning73%

Recent Tests

GPT-4o prompt optimization
2 hours ago
Claude-3.5 tool calling
6 hours ago
RAG chunk size A/B test
12 hours ago

Evaluation Framework

1,247 total tests

Automated testing pipeline with custom metrics, human feedback loops, and A/B testing. Tracks accuracy, hallucination rates, tool usage effectiveness, and user satisfaction scores across different model versions and prompt templates.

Workspace

About

I build AI products end‑to‑end: data ingestion, retrieval, prompt/tooling, evals, and production UI. Pragmatic about latency, cost, and safety — with strong attention to developer experience.

8+

Years

120+

Projects

50+

Clients

Currently learning

Structured outputs, memory architectures, and low‑latency tool use with vLLM + GPU batching.

Let’s collaborate

Let's Connect

Have a project in mind, a question, or just want to say hello? I'd love to hear from you. Fill out the form, and I'll get back to you as soon as possible.

Send me a message