Looking for a job

faras siddiqui

@farassiddiqui

Message

ML engineer at CVision building hardware-aware AI runtimes and a neuromorphic compiler delivered six months early.

Pakistan

Message

At CVision, I built TALON’s compile path for Type 1 Compute, including graph partitioning across neuromorphic cores, placement and routing, CPU simulation, and latency and energy profiling. I also delivered an hls4ml FPGA backend about six months ahead of schedule.

At Noah Labs, I built GPU-cluster scheduling and vLLM inference orchestration for an AI coding IDE piloted with Lockheed Martin and Pegasus Airlines. I moved agent workloads from commercial APIs to open-weight models on NVIDIA GPUs, with task-aware evaluation and routing and autoscaling from one GPU to the full cluster.

Experience

Work history, roles, and key accomplishments

Current

Founding Engineer

Current

CVision

Feb 2025 - Present (1 year 9 months)

Built the TALON compile path for Type 1 Compute, including graph partitioning, core placement, and FPGA backend. Delivered production AI systems for industrial and SME partners, generating $150K+ in first-year revenue.

Model To Hardware Compiler Neuromorphic Computing Hls4ml FPGA Python PyTorch CI CD Docker Kubernetes

AI Systems Engineer

Noah Labs

Feb 2026 - Oct 2026 (8 months)

Built GPU-cluster scheduling and inference orchestration with vLLM across NVIDIA A100, H100, GH200, and B200. Moved agent workloads off Anthropic and OpenAI APIs onto open-weight models, reducing costs by 50-65%.

VLLM GPU Cluster Scheduling KV Caching Model Evaluation Routing A100 NVIDIA H100 GH200 NVIDIA B200 AutoScaling Cost Optimization

Deep Learning Research Intern

Optical Networks and Technologies Lab, NUST

Jun 2024 - Sep 2024 (3 months)

Implemented and trained transformer and GAN models for optical-signal classification and anomaly detection. Explored attention mechanisms and evaluated performance on optical-network data.

Transformers Gan PyTorch Optical Signal Processing Anomaly Detection