Curriculum Vitae

Syed Muhammad Hussain

Karachi, Pakistan
P: +92 3453887230
E: syed.hussain@beam.ai
LinkedIn | GitHub | Google Scholar

Education

Habib University
Bachelor of Science | Aug 2020 - April 2024
Major in Computer Science | Minor in Electrical and Computer Engineering
Relevant Coursework: Deep Learning, Software Engineering, Machine Learning, Algorithms, Data Science

Work Experience

BEAM AI
Machine Learning Engineer I | May 2025 - Present
Associate Machine Learning Engineer | May 2024 - May 2025
New York, United States · Hybrid

At Beam AI, I contribute to the development and deployment of AI-driven systems designed to automate complex workflows across enterprise use cases. My role spans research, implementation, and optimization of large-scale language model applications.

Key responsibilities and contributions include:

• Custom LLM Evaluation: Developed and applied bespoke evaluation metrics to systematically assess large language model (LLM) outputs for hallucinations, factual accuracy, and task-specific alignment.
• Deterministic Code Interpreter: Designed a deterministic code interpreter to ensure controlled and reliable execution of model-generated code, improving reproducibility and safety in automated pipelines.
• Guardrails for LLM Outputs: Integrated and customized guardrails frameworks to enforce output constraints, ensuring model responses adhere to compliance, ethical, and performance standards.
• Meta Prompting for Complex Tasks: Engineered multi-layered meta prompting strategies to guide LLMs through complex, multi-step tasks with enhanced contextual awareness and reduced error propagation.
• Retrieval-Augmented Generation (RAG): Built and fine-tuned RAG pipelines to support real-time contextual generation, combining external knowledge bases with generative reasoning.

This role allows me to work at the intersection of machine learning engineering, prompt design, and evaluation science, pushing the boundaries of what LLMs can achieve in real-world systems.

LangChain
Open Source Contributor | Mar 2025 - Present

Actively contributing to the LangChain ecosystem, particularly the langchain-google integration. My contributions focus on improving documentation, ensuring compatibility with updated dependencies, and addressing community-reported issues.

Key Contributions:

• Enhanced Documentation: Authored a comprehensive README for the langchain-google-vertexai integration to improve developer onboarding and clarity. PR #934
• Code Maintenance: Fixed compatibility issues with Pydantic v2.11+ by updating deprecated model_fields access. PR #931
• Community Engagement: Provided input and solutions in multiple GitHub issues including #948, #936, #907, and #945

University of South Australia - Empathic Computing Laboratory
Research Intern – Cognitive Systems & Machine Learning | Sep 2024 - Mar 2025
Australia · Remote

Contributing to cutting-edge research at the intersection of cognitive neuroscience and machine learning. My work focuses on the analysis of EEG and physiological signals to detect and model user preferences and cognitive states.

Key contributions:

• Developed pipelines for multimodal data preprocessing and feature extraction from EEG and biosignals.
• Conducted statistical and time-series analysis to identify signal patterns related to user intent and preference.
• Designed and implemented machine learning models for classifying user responses in human-computer interaction experiments.

This role combines applied ML, neuroscience, and experimental design, contributing to next-gen adaptive interfaces that better understand human intent.

Habib University
Undergraduate Researcher – Computer Vision & Deep Learning | May 2023 - Jan 2024
Karachi, Pakistan · On-site

Led a research initiative focused on camouflaged object detection, a critical challenge in military surveillance, wildlife monitoring, and autonomous systems.

Project highlights:

• Investigated the limitations of conventional object detection models in recognizing low-contrast or concealed objects.
• Pioneered a synthetic data generation pipeline using Generative Adversarial Networks (GANs) to simulate complex camouflage patterns across diverse scenes.
• Designed and trained deep learning models with custom datasets, significantly improving detection performance in challenging visual environments.
• Published findings and insights with relevance to robust computer vision systems for real-world applications.

Folio3 Pakistan
Machine Learning Engineer – Computer Vision | Jul 2023 - Sep 2023
Karachi, Pakistan · Hybrid

Worked on real-world computer vision projects involving object detection and image segmentation, applying deep learning to develop scalable solutions for industrial use cases.

Key achievements:

• Developed and fine-tuned object detection and instance segmentation models using state-of-the-art architectures (e.g., YOLO, Mask R-CNN).
• Optimized model performance through data augmentation, hyperparameter tuning, and inference pipeline refinement.
• Gained hands-on experience with labeling workflows, model deployment pipelines, and real-time vision systems.
• Collaborated cross-functionally with engineers and product teams to align ML deliverables with business needs.

Habib University
Project Assistant | Jun 2022 - Sep 2022
Karachi, Pakistan · On-site

Contributed to the design and development of an Arduino-controlled CNC dividing machine, built to enhance the precision and automation of milling operations.

Key contributions:

• Hardware-Software Integration: Supported the setup and synchronization of key components, including stepper motors, drivers, sensors, and keypad interfaces with the Arduino microcontroller.
• System Calibration & Testing: Conducted systematic testing and troubleshooting to ensure reliable angular rotation, enabling precise hole placements at user-defined intervals.
• Automation for Precision: Eliminated manual rotation errors by implementing a keyboard-input angle control system.
• Feedback-Control Enhancement: Contributed to developing a feedback control loop to minimize rotational error.
• Documentation & Analysis: Assisted in documenting the system design, operational procedures, and performance data.

Projects

1. Intelligent Enterprise Search with RAG

Application: AI-driven knowledge discovery for enterprises, enabling instant access to critical information.

• Developed a high-performance Retrieval-Augmented Generation (RAG) pipeline for intelligent document search.
• Engineered a multi-stage indexing system with semantic chunking and vectorized retrieval using Qdrant & PGVector.
• Integrated LangChain, LlamaIndex, and LLMs (OpenAI, Anthropic, Mistral) to enhance domain-specific information retrieval.
• Applied advanced prompt engineering (CoT, Few-Shot Learning), improving search accuracy by 40%.
• Deployed on Azure for scalable, real-time document retrieval with minimal latency.

2. Scalable MLOps for AI Model Deployment & Monitoring

Application: Automating AI model deployment and lifecycle management for enterprise solutions.

• Designed an end-to-end MLOps pipeline with Docker & MLflow, ensuring seamless model deployment and version control.
• Built an AI evaluation framework using Guardrails & Geval for real-time assessment of hallucination rates, reasoning accuracy, and robustness.
• Integrated LLMs into production workflows via Azure AI Services, automating data-driven decision-making.
• Improved system reliability and responsiveness through automated monitoring, anomaly detection, and performance optimization.

3. AI-Driven Document Intelligence & Visual Reasoning

Application: Automating document processing and analysis for legal, financial, and research industries.

• Developed a multimodal AI system for intelligent extraction and interpretation of structured and unstructured documents.
• Automated parsing and semantic analysis of PDFs, CSVs, DOCX, and PPTs using Azure & AWS AI services.
• Implemented Visual Question Answering (VQA) to enable contextual document navigation and reasoning.
• Strengthened document understanding and interpretation using customized models architectures in PyTorch, TensorFlow, and Keras.

4. Autonomous AI Agent for Workflow Optimization

Application: Intelligent task automation for enterprise operations and decision-making.

• Engineered an AI agent capable of autonomous task planning, execution, and adaptive decision-making.
• Designed a multi-agent system leveraging LangChain & LlamaIndex for contextual awareness and long-term memory.
• Applied structured reasoning (CoT & ToT prompting) to optimize multi-step workflows.
• Integrated OpenAI, Anthropic, and Mistral APIs, enhancing adaptability to dynamic task requirements.
• Minimized the need for manual oversight by automating complex workflows, improving operational efficiency.

Publications

• Navigating the Maze Environment in ROS-2: An Experimental Comparison of Global and Local Planners for Dynamic Trajectory Planning in Mobile Robots
  S.M. Hussain, Syed Muhammad Daniyal Murtaza Zaidi, Ahmed Atif, Laiba Ahmed, Dr. Qasim Pasta, Dr. Basit Memon.
  2024 50th IEEE Industrial Electronics Conference (IECON). Chicago, USA.
  (Acceptance Letter)

• Advancing Real-Time Camouflaged Object Detection with YOLOv8
  S.M. Hussain, Afsah Hyder, Samiya Ali Zaidi, S.M. Ali Rizvi, Dr. Muhammad Farhan.
  2024 6th International Conference on Advances in Computer Vision, Image and Virtualization. Sanya, China.
  (Acceptance Letter)

• Enhanced Camouflaged Object Detection for Agricultural Pest Management: Insights from Unified Benchmark Dataset Analysis
  S.M. Hussain, Afsah Hyder, Samiya Ali Zaidi, S.M. Ali Rizvi, Dr. Muhammad Farhan.
  20th International Bhurban Conference on Applied Sciences and Technology IBCAST 23.

• In Automated Video Summarization for Suspicious Event Detection in Surveillance Systems: A Pipeline Approach
  S.M. Hussain, Azeem Haider, Affan Habib, Dr. M.Farhan.
  20th International Bhurban Conference on Applied Sciences and Technology IBCAST 23. Islamabad, Pakistan.

• Integrated Ensemble Learning into Remote Health Monitoring for accurate prediction of oral diseases
  S.M. Hussain, Samiya Ali Zaidi, Afsah Hyder, Dr. M.Mobeen Movania.
  25th International Multitopic Conference INMIC 23. Lahore, Pakistan.

• A Machine Learning Approach to Arabic Phoneme Classification through Ensemble Techniques
  S.M. Hussain, Muhammad Najeeb Jilani, Munim Ul Haq, Dr. Muhammad Shahid Shaikh.
  5th International Conference on Advancements in Computational Sciences. Lahore, Pakistan.

Extra-Curriculars

Medium Blogs
Selected Articles:

• Single-Stage Object Detectors: A Fast Approach to Object Detection
• Usability Evaluation of Financial Banking and Ride-Hailing Applications
• The AI Trees (Design Fiction)
• Early Childhood Development And Interaction With Technology In Pakistan
• Understanding Human-Computer Interaction

Texas A&M: Invent For The Planet
Team Leader:

• Led a team to address global challenges in a 48-hour innovation event, focusing on teamwork and collective action.

Additional Information

Technical Skills:
PyTorch, TensorFlow, Keras, Pandas, NumPy, Data Science, Machine Learning, Deep Learning, Natural Language Processing (NLP), Retrieval-Augmented Generation (RAG), AI Agents, Computer Vision, MLOps, Cloud Computing, Prompt Engineering, Model Evaluation, Data Engineering, Algorithms & Design, SQL, DevOps.

Honors & Awards:

• Awarded a fully funded 4-year scholarship through the Habib University TOPS program.
• Selected as a Research Intern at the Empathic Computing Lab, a leading AI research lab in ML and HCI, as the only research intern from my Pakistan region.