Haris Ahmed
Contact
’26
Available · May 2026Lahore · PKT --:--

Haris Ahmed.

AI engineer, building quietly.

Prototypes impress. Production systems pay. I build the latter.

View work
Get in touch
/ Live · agent runRAG · 6 tools
agent.trace · run #0247● live
00·router.classify("hire ai engineer")→ intent=portfolio42ms
01·retriever.search(k=8 · rerank)→ Recall@4 = 0.91191ms
02·llm.stream(gpt-4o · 3.2k ctx)→ 84 tok/s · $0.0019612ms
03·trace.log(ok)→ total 851ms6ms
total · 851mscached · cost $0.0019
Stack
Python/TypeScript/Next.js/FastAPI/LangChain/RAG/Postgres/Stripe/Docker/Remotion/Python/TypeScript/Next.js/FastAPI/LangChain/RAG/Postgres/Stripe/Docker/Remotion/
01 / AboutSubject — Haris Ahmed

I build intelligent systems that survive production.

Most AI projects fail the same way. A prototype impresses in a demo, then collapses under real traffic, real edge cases, and real costs. I am the engineer you bring in before that happens.

I have shipped real-time adversarial AI platforms, RAG-powered engines with natural language control, automated billing systems with tiered access and credit workflows, and media pipelines that cut asset turnaround by 80 percent. Every system I build is designed to reduce your inference costs, cut manual effort, and scale without rewrites.

I work across the full stack: Python, FastAPI, LangChain, LLM orchestration, Pinecone, PostgreSQL, Redis, and Next.js. I do not just write code. I make architectural decisions that save you money at scale through smarter caching, leaner prompts, async queues, and pipelines that do not break under pressure.

You get one engineer who owns the outcome end to end. No handoffs. No gaps between frontend, backend, and AI layers. Faster delivery, lower overhead, and a system built to last.

Role
Full-Stack · AI Engineer
Based
Pakistan
Status
Available
Timezone
UTC+5 · PKT
02 / StackTools of the trade

The toolkit that ships production.

AI / ML

07
Prompt EngineeringRAGVector EmbeddingsLLM OrchestrationAgentic WorkflowsFine-tuning (LoRA/QLoRA)Multi-Agent Systems

Languages

06
PythonSQLJavaScriptReactTypeScriptNoSQL

Frameworks

05
FastAPILangChainNext.jsCrewAINode.js

Stores

07
PostgreSQLRedisQdrantFAISSFirebasePineconeMongoDB

Infra

05
DockerGitRailwayVercelGitHub Actions

Tools

05
Claude CodeCodexCursorOpenAPI/SwaggerStripe API
Hero
About
Stack
Scroll
PrinciplesClean architectureDRY · SOLID · KISSAI-first designPerformance obsessedProduction mindset
01+
Year shipping product
10+
Projects in production
06
Stacks in active use
∞
Curiosity budget
03 / Selected work2024 — 2026

Selected work, shipped.

A curated archive of engineered systems built shipped and production ready not prototypes

/01 · FEATURED

HardTalk: AI Rehearsal Platform for High-Stakes Conversations

Real-time voice rehearsal for investor pitches, board briefings, sales calls, and media interviews. Up to three AI personas interrupt, push back, and escalate pressure over a single Gemini Live WebSocket. Post-session analysis scores performance across six skill axes and prescribes targeted drills.

Voice latency
sub 300ms
Gemini Live response budget
Personas per room
3 to 6
concurrent AI participants
Skill axes scored
6
rubric per session
TypeScriptSupabasePostgresDeno Edge FunctionsGemini LiveVertex AIGrok / X.AIStripeVitestReact Query
Year2026
RoleFull-stack engineer
StatusShipped
ScopeFull-stack
LiveCase Study
/02 · FEATURED

Scout: Centralized Intelligence Data Platform

A centralized ingestion platform that orchestrates Apify actors, Firecrawl, Crawlee, Playwright, and Scrapling behind a single unified API. Discovers sources through an LLM ReAct loop, extracts structured records with content-anchored prompts to eliminate hallucinations, tracks cost per job, and enforces full project-level isolation. Ships with a Next.js admin dashboard.

TypeScriptFastifyNode.jsPrismaPostgreSQLNext.jsApifyFirecrawlZodRedisRailway
Year2026
RoleLead Engineer
StatusIn motion
ScopeFull-stack
Case Study
/03 · FEATURED

Haris. AI Powered Developer Portfolio

A full-stack developer portfolio with a RAG-grounded AI chat assistant, an admin dashboard, and LLM-powered resume tailoring. Next.js 16 plus FastAPI, two containers, one SQLite file.

Backend routes
57
Across 10 FastAPI routers covering public, chat, and admin surfaces.
Backend tests
116
Pytest suite covering SSE chat, guard branches, quota, memory, and admin CRUD.
Containers
2
Frontend plus backend; SQLite file and JSON RAG live on a mounted volume.
Next.jsReactTypeScriptTailwind CSS v4Framer MotionReact Three Fiberthree.jsLenisZodshadcn/uiFastAPIPythonuvaiosqliteSQLitenumpyOpenRouterDeepSeeksse-starletteWeasyPrintpdfplumberFernetDocker ComposeMake
Year2026
RoleLead Engineer
StatusIn motion
ScopeFull-stack
LiveCase Study
/04 · FEATURED

Inflectiv: SaaS API Gateway and Monetization Engine

Designed and built the credit-based SaaS monetization engine powering Inflectiv, a decentralized data infrastructure platform that structures, tokenizes, and monetizes datasets for AI agents. Delivered a scalable external API gateway with custom rate limiting, automated OpenAPI/Swagger documentation, multi-tier access control across Free, Basic, and Pro plans, and an end-to-end Stripe-powered credit billing system.

Access tiers
3
Free, Basic, and Pro with per-tier feature gating
Billing model
Per-query
Stripe-powered credit system with automated top-up
API coverage
Internal + External
Credit system spans both internal and public-facing APIs
TypeScriptFastApiPrismaPostgreSQLStripeRedisSwaggerREST API
Year2025
RoleAssociate AI Engineer
StatusShipped
ScopeBackend
LiveCase Study
/05 · FEATURED

LogoForge

An AI logo generator that turns a short prompt into a transparent PNG and a clean SVG, with a chat mode that asks a few questions to sharpen the brief before drawing.

Apps
2
frontend + API
Output formats
PNG + SVG
raster and vector
Status states
7
pipeline stages
Next.jsTypeScriptZodNextAuth.jsPostgreSQLOpenRouterVTracerDocker
Year2025
RoleLead Engineer
StatusShipped
ScopeFull-stack
Case Study
/06 · FEATURED

FitMeal Planner

An AI powered weekly meal planning app that builds a 7 day plan from your goals, diet, and allergies, with recipes, shopping lists, and reminders on web, Android, and iOS.

ReactTypeScriptTailwind CSSTanStack QueryFirebaseOpenAI APICapacitorAndroidiOS
Year2025
RoleLead Engineer
StatusShipped
ScopeFull-stack
Case Study
/01 · FEATURED

HardTalk: AI Rehearsal Platform for High-Stakes Conversations

Real-time voice rehearsal for investor pitches, board briefings, sales calls, and media interviews. Up to three AI personas interrupt, push back, and escalate pressure over a single Gemini Live WebSocket. Post-session analysis scores performance across six skill axes and prescribes targeted drills.

Voice latency
sub 300ms
Gemini Live response budget
Personas per room
3 to 6
concurrent AI participants
Skill axes scored
6
rubric per session
TypeScriptSupabasePostgresDeno Edge FunctionsGemini LiveVertex AIGrok / X.AIStripeVitestReact Query
Year2026
RoleFull-stack engineer
StatusShipped
ScopeFull-stack
LiveCase Study
/02 · FEATURED

Scout: Centralized Intelligence Data Platform

A centralized ingestion platform that orchestrates Apify actors, Firecrawl, Crawlee, Playwright, and Scrapling behind a single unified API. Discovers sources through an LLM ReAct loop, extracts structured records with content-anchored prompts to eliminate hallucinations, tracks cost per job, and enforces full project-level isolation. Ships with a Next.js admin dashboard.

TypeScriptFastifyNode.jsPrismaPostgreSQLNext.jsApifyFirecrawlZodRedisRailway
Year2026
RoleLead Engineer
StatusIn motion
ScopeFull-stack
Case Study
01/06scroll to flip
More in the archive
View all projects
Every repo · shipped, retired, & in motion
04 / PathA chronology of craft

Where I've been, in order.

Aug 2025 — Present
Work

Associate AI Engineer

Vanar

Building production SaaS products: Inflectiv (API gateway with Stripe billing & multi-tier access control), Hard Talk (AI rehearsal platform with adversarial LLM personas), Fitmeal Planner (RAG-powered nutrition engine), and automated video generation pipelines with Minimax & Remotion.

PythonFastAPIStripeLLMsRAGRemotion
2021 — 2025
Study

Bachelor of Science in Computer Science

National University of Modern Languages, Lahore

Graduated with a 3.4 CGPA. Focused on algorithms, distributed systems, and machine learning. Built multiple capstone projects involving NLP and AI applications.

05 / Contact

Let's build something real.

harisahmed510.00@gmail.com
GitHubLinkedInEmail
Haris Ahmed

AI engineer building intelligent systems that survive production. Available for roles & contract work.

Back to top
IndexAboutStackWorkPathContact
ElsewhereGitHubLinkedInEmail
© 2026 Haris Ahmed · All rights reservedAI systems that actually scale.
haris-ai.session
Live
Haris

Haris AI

Retrieval-augmented · Always on

Hi, I'm Haris's AI. Ask me about his work, his stack, or how to reach him. I'll get you straight to the answer.

Try asking
Enter to send · Shift+Enter for newline