a
arhamakmalkhan

Arham Akmal

@arhamakmalkhan

Making AI Work for You! Simple, Smart, and Effective

Pakistan
Anglais, Ourdou, Pachto
Certaines informations sont présentées en anglais.
À propos de moi
I am an AI Engineer specializing in production ready multi-agent systems and Retrieval Augmented Generation (RAG) platforms. I help enterprises automate complex data pipelines and orchestrate scalable AI search. With a proven track record of reducing LLM inference costs by 70% and handling high-throughput asynchronous data ingestion (up to 10,000 rows/second), I deliver secure, high-performance solutions. My core tech stack includes Async Python, FastAPI, LangGraph, Qdrant, and PostgreSQL. Let’s build reliable AI systems that drive measurable business outcomes.... Plus d’infos

Compétences

a
arhamakmalkhan
Arham Akmal
hors ligne • 
Temps de réponse moyen de 1 heure

Voir mes services

Développement de chatbots d'IA
I will build and integrate ai chatbots
De25 $US / projetPlus de détails
Mise en œuvre et déploiement de l’IA
I will build custom multi agent ai systems and production rag pipelines
De30 $US / projetPlus de détails

Portfolio

Expérience professionnelle

Oak_Street Technologies

AI Engineer

Oak Street Technologies

Jan 2025 - Present • 1 yr 8 mos

Architected a Custom Multi-Agent Conversational System: Engineered a production-grade LangGraph-based rea- soning engine handling 100+ daily queries, with response times of 15 to 20 seconds depending on query complexity. Implemented SQL execution nodes with retry logic and strict guardrails, reducing hallucination rates by 70%. • Built an Asynchronous Data Ingestion Pipeline: Engineered a robust job queue system capable of handling 10,000 rows per second, processing structured reports into PostgreSQL via Azure Blob storage and dynamic background work- ers. • Spearheaded Vector Infrastructure Optimization: Benchmarked Qdrant, Weaviate, and ChromaDB for hybrid search performance and scale. Selected and deployed Qdrant to maximize query throughput and minimize latency for produc- tion RAG workloads. • Engineered Dynamic Context & Memory: Implemented dynamic prompt injection to supply agents with real-time schema structures and integrated Qdrant to establish scalable, long-term conversational memory. • Deployed Production LLMOps & API Security: Integrated Langfuse for end-to-end LLM observability and deploy- ment optimization, successfully reducing inference costs by almost 70%. Secured the API infrastructure with strict rate- limiting, maintaining 99.9% uptime.