Talha Yousaf Saeed

Software & AI Systems Engineer specializing in production-grade backend and LLM orchestration.

Based in: Lahore, PakistanCurrently: Software & AI Consultant, Branqor (independent consultancy)Experience: 2 years

Hire me

Experienced Software and AI Systems Engineer with a strong background in building production backend systems for fintech, compliance, and B2B SaaS products. He has a proven track record of independently shipping multi-tenant AI platforms and automating complex business workflows using Python, Go, and Next.js.

Experience

Software & AI Consultant

Branqor (independent consultancy)Remote / US Clientfreelance2026 – Present

Summary

Engaged by a US client to design and deliver AI-driven workflow automation

What I did

  • Build workflows connecting CRM, cloud and financial data sources into a decision-support layer for the operations team.
  • Automated 3 recurring workflows, reducing manual processing time by an estimated 20%; own requirements, architecture, delivery and client communication.

Results

  • Automated 3 recurring business workflows for a US client, reducing manual processing time by an estimated 20%.
AICRMCloudFinancial Data Sources

Software Engineer

Programmers ForceworkAug 2024 – 2026

Summary

Backend and AI engineering across Shufti Pro, AML Watcher and Media Watcher

What I did

  • Platform: Built Python and Laravel REST services covering authentication, identity verification, screening and data processing.
  • Authentication: Implemented OAuth 2.0 (Google, LinkedIn, HubSpot, Apple), JWT sessions and two-factor authentication.
  • AI integration: Integrated an MCP client/server layer for tool-based communication between internal services and LLM pipelines.
  • Shufti Pro: Built backend features for identity verification and real-time AML screening in a regulated fintech product.
  • AML Watcher: Built sanctions and watchlist screening services with ingestion pipelines for multiple data sources.
  • Media Watcher: Built Flask, MongoDB and Elasticsearch infrastructure for news and social intelligence, with semantic search, Twitter/Reddit/YouTube integrations and Stripe billing.
  • Managed screenings for IDs, passports, and facial verification.
  • Optimized systems for low false positive rates.
  • Converted an acquired product's codebase to the Gin framework for integration into Shufti Pro.
  • Implemented a transaction monitoring system where clients define rules for models to flag irregular activities.

Results

  • Delivered backend features across 3 production products (Shufti Pro, AML Watcher, Media Watcher) in regulated fintech and compliance domains.
  • Implemented OAuth 2.0 for 4 providers (Google, LinkedIn, HubSpot, Apple) and built an MCP security proxy supporting 4 LLM platforms.
  • Handled roughly 30,000 - 40,000 requests per day for ID, passport, and facial verification screenings.
  • Maintained request resolution targets of under 30 - 45 seconds.
PythonLaravelREST APIsOAuth 2.0JWT2FAMCPLLMFlaskMongoDBElasticsearchTwitter APIReddit APIYouTube APIStripeGoGin

Skills

Technical

OAuth 2.0
JWT
PostgreSQLPostgreSQL
REST APIs
PythonPython
LLM orchestration
FastAPIFastAPI
RAG
TypeScriptTypeScript
GoGo
DockerDocker
FlaskFlask
LaravelLaravel
pgvector
MongoDBMongoDB
Elasticsearch
RedisRedis
LangChain
Next.jsNext.js
ReactReact
Gin
Reranking
Structured Decoding
JSON Schema
Ollama
Pydantic
Qwen
Quantization
GPTQ

Languages

Urdu
English

Projects

Swiv AI

Solo Developerpersonal
Multi-tenant B2B SaaS for competitor monitoring, pricing and website change detection, and market research. Backend: Async FastAPI backend on PostgreSQL, Redis, SQLAlchemy 2 and Alembic, separating APIs, domain logic and background workers. AI pipelines: Research pipelines combining Firecrawl, Exa and LLMs for competitor analysis, significance scoring and summarization. RAG assistant: Streaming Ask assistant using SSE and tool calling, with persistent threads and cited answers over pgvector search. Security & billing: JWT authentication, RBAC, entitlement-based access and token/cost usage metering. Frontend: Shipped the Next.js and TypeScript web app, marketing site and streaming chat interface.
FastAPIPostgreSQLRedisSQLAlchemy 2AlembicFirecrawlExaLLMsSSEpgvectorJWTNext.jsTypeScript
Visit the site

Invigilate

Developerpersonal
Built a transparent JSON-RPC 2.0 proxy between AI agents and MCP servers applying governance controls without changing agent or server code. Implemented allow, block and notify rules by tool name, risk level and arguments (destructive SQL, shell commands, file deletion, unsafe writes). Added schema sanitization for AWS Bedrock, Azure OpenAI, Google Vertex AI and Anthropic, Redis caching, audit logs, rate limiting and encrypted credential storage.
FastAPISQLAlchemyPostgreSQLRedisAlembicPythonJSON-RPC 2.0AWS BedrockAzure OpenAIGoogle Vertex AIAnthropicOpenAI API

Local Agentic Tasks with Ollama

personal
Hosted LLM models locally on a MacBook to perform small agentic tasks in the background.
OllamaQwen

Education

BSc Computer Science

Riphah International UniversityComputer ScienceGPA 3.392020 – 2024

Recognition

Certifications

AWS Cloud Architecting

AWS Academy

Generative AI

Google

Blockchain Systems

UC Irvine

DevOps, DataOps and MLOps

Duke University
Hire TalhaReach out about a role, a contract or a conversation.For recruiters

Talha's twin is AI, it can make mistakes.

More people on Kanvis

Browse more peopleCreate your own Kanvis