Senior AI Machine Learning Engineer

Techsa; Egypt (PA); Hybrid; USD 120,000 - 190,000; Full time

Job summary

Techsa seeks an experienced ML/AI engineer to own end-to-end systems, from data pipelines to deployment of LLM-powered applications. You will build custom pipelines, manage prompts, and implement autonomous workflows using multi-agent orchestration technologies.

The role involves developing and evaluating RAG pipelines, optimizing embeddings and re-ranking, and deploying on-premise inference infrastructure with vLLM or Ollama. Collaboration with Data Engineering is key.

Qualifications

• 3+ years of hands-on ML/AI engineering with end-to-end ownership.
• Production experience building LLM-powered applications.
• Hands-on experience with agent orchestration in production.
• Production RAG experience with evaluation metrics, hybrid search, and re-ranking.
• Experience building ML models for churn, propensity, LTV, or segmentation.
• Hands-on experience with data pipelines: Spark for batch, Flink or Kafka Streams for real-time.
• Experience with vector databases at scale (OpenSearch k-NN, Qdrant, Milvus).
• Real-time ML inference experience at 1,000+ QPS.

Responsibilities

• Own ML/AI systems end-to-end from data to deployment.
• Build LLM-powered applications with custom pipelines and prompt management.
• Implement multi-agent orchestration for autonomous workflows.
• Build and optimize RAG pipelines with embedding and re-ranking strategies.
• Deploy and manage LLM inference infrastructure (on-premise).
• Design AI operators for visual low-code canvases and gateways.
• Collaborate with Data Engineering and Platform teams to integrate ML systems.

Skills

End-to-end ML ownership

LLM-powered applications

Agent orchestration in production

RAG pipelines evaluation

ML model development (churn/propensity

Real-time data pipelines (Spark/Flink)

Vector databases at scale

Real-time inference 1k+ QPS

MLOps practices

On-premise inference (vLLM/Ollama)

Tools

LangGraph

CrewAI

AutoGen

LlamaIndex

vLLM

Ollama

Spark

Flink

Kafka Streams

OpenSearch k-NN

Qdrant

Milvus

Job description

• - Own ML/AI systems end-to-end: data pipelines, modeltraining, serving infrastructure, monitoring, and iteration
• - Build LLM-powered applications with custom pipelines,prompt management, evaluation, and optimization
• - Implement multi-agent orchestration systems usingLangGraph, CrewAI, or AutoGen for autonomous workflows
• - Build and optimize RAG pipelines using LlamaIndex withchunking strategies, embedding selection, re-ranking, and evaluation
• - Deploy and manage LLM inference infrastructure using vLLMor Ollama for on-premise sovereign deployments
• - Build traditional ML scoring models: churn prediction,propensity scoring, LTV estimation, next-best-action
• - Design and build feature pipelines using Apache Flink(streaming) and Spark (batch) for real-time and batch ML
• - Implement MLOps practices: model versioning, registry,drift monitoring, A/B testing, and staged rollouts
• - Design and implement AI operators for visual low-codecanvas (LLM Gateway, RAG Pipeline, Intent Classifier)
• - Optimize ML inference for latency and throughput at scale(10K+ QPS)
• - Collaborate with Data Engineering and Platform teams tointegrate ML systems with data infrastructure

Requirements

• - 3+ years of hands-on ML/AI engineering with demonstratedend-to-end system ownership
• - Production experience building LLM-powered applications(not just API consumption)
• - Hands-on experience with agent orchestration: LangGraph,CrewAI, or AutoGen in production
• - Production RAG experience with evaluation metrics, hybridsearch, and re-ranking strategies
• - Experience building ML models: churn, propensity, LTV,segmentation, recommendation systems
• - Hands-on experience with data pipelines: Spark for batch,Flink or Kafka Streams for real-time
• - Experience with vector databases at scale: OpenSearchk-NN, Qdrant, or Milvus
• - Real-time ML inference experience at 1,000+ QPS

Good to Have

• - Experience at AI-first companies or building AI/MLplatforms from scratch
• - Telco or enterprise data platform background
• - Experience with LLM fine-tuning: LoRA, QLoRA, PEFTtechniques
• - Experience with embedding models: sentence-transformers,fine-tuning for domain
• - Kubernetes for ML workload orchestration and GPUscheduling
• - Knowledge of PII detection (Presidio) and LLM guardrails(NeMo Guardrails)