Agentic AI
Systems&Engineer
Specializing in autonomous multi-agent pipelines (LangGraph), production RAG retrieval, and resilient FastAPI backends.
Architecting enterprise-grade autonomous agent systems, multi-hop RAG pipelines, and deterministic LLM evaluation frameworks. Focused on lowering compute latency, scaling async backends, and enforcing reliable agentic execution loops.
Featured Architectures
Custom agentic graphs, hybrid vector indexing, and deterministic LLM evaluation engines built for enterprise uptime.
Autonomous Multi-Agent Workflow Engine
A distributed, stateful multi-agent execution pipeline built on LangGraph that coordinates specialized sub-agents with dynamic tool-calling, fallback recovery, and cyclical state checkpoints.
Reduced manual business process runtime by 20% while eliminating state desynchronization.
Hybrid Production RAG Pipeline
An enterprise-scale Retrieval-Augmented Generation engine combining dense semantic vector search with sparse keyword indexing, custom dynamic chunking, and cross-encoder re-ranking.
Maintains sub-500ms P99 retrieval latency across 500k+ enterprise documents with zero context leakage.
Enterprise LLM Evaluation Platform
A comprehensive LLM evaluation and continuous observability platform utilizing automated edge-case test suites, prompt regression checks, RLHF telemetry, and token cache optimization.
Cut recurring compute costs by 25% through prompt token caching while raising prompt instruction adherence by 15%.
Technical Stack
Production-grade tooling built around deterministic agent graphs, hybrid retrieval indexes, and high-concurrency async services.
Agentic AI & LLMs
Autonomous Reasoning & Multi-Agent Graphs
Stateful cyclic agent workflows
Chain composition & custom loaders
Hierarchical sub-agent swarms
Deterministic automated test suites
Open-weights model fine-tuning
ReAct, Tree-of-Thoughts & Reflection
Data & Vector Search
High-Dimension Indexing & Retrieval
Backend & Core
Resilient Asynchronous Services
Engineering Practices
Production Observability & Optimization
Multi-stage context retrieval
Production tracing, cost & token monitoring
Few-shot optimization & structured outputs
Caching strategies & compute reduction
Ranking datasets & instruction alignment
Async batching & streaming SSE
Engineering Experience
High-impact engineering roles optimizing agent latency, training eval benchmarks, and scaling backend services.
* Hover or click rows for inverted contrast viewAB {Ark} Solutions
Associate AI EngineerEngineering core enterprise agentic pipelines and production RAG services. Designing multi-agent decision systems with stateful cyclic graphs and deep observability.
Engineered multi-agent orchestration pipelines utilizing LangGraph, enabling autonomous business task execution with full state recovery.
Constructed hybrid RAG retrieval systems integrating Weaviate and custom re-ranking models for high-precision knowledge queries.
Integrated Langfuse telemetry across all LLM inference endpoints to track latency, token drift, and model hallucinations in production.
Turing
Software Developer (LLM Evaluation)Spearheaded technical LLM evaluation benchmarks and RLHF alignment workflows across complex code generation and reasoning tasks.
Viztera Solutions
Software DeveloperDeveloped high-throughput Python backend microservices, optimized database queries, and implemented secure asynchronous REST APIs.
Education & Research
Rigorous background blending Control Systems & Autonomous Feedback with Advanced Machine Learning and Vector Architectures.
Sohaib Sultan
Agentic AI Systems Engineer
Combining control theory rigor with stateful agent graphs to build resilient AI systems that degrade gracefully and recover deterministically.
MS Data Science
Information Technology University (ITU)
BS Mechatronics & Control Engineering
University of Engineering & Technology (UET)
Let's build reliable autonomous systems.
Looking for an Agentic AI Systems Engineer to architect production LangGraph workflows, hybrid RAG retrievers, or high-throughput FastAPI backends?