Usman Jalil

Quanteam: Contributed to the development of core services for an always-on Agentic AI platform enabling users to create custom AI agents, integrated across 21 industries, 123+ roles, and a skills taxonomy of over 22k+ capabilities Facilitated seamless deployment of agents across multiple channels, including Slack, Discord, Telegram, and WhatsApp, allowing them to perform daily routine tasks autonomously Built an enterprise-grade RAG connector with RBAC to manage and query niche, domain-specific documents using the existing suite of agents

Software Engineer at True Meridian PVT LTD (2024-06 – Present)

• Quanteam: Contributed to the development of core services for an always-on Agentic AI platform enabling users to create custom AI agents, integrated across 21 industries, 123+ roles, and a skills taxonomy of over 22k+ capabilities
• Facilitated seamless deployment of agents across multiple channels, including Slack, Discord, Telegram, and WhatsApp, allowing them to perform daily routine tasks autonomously
• Built an enterprise-grade RAG connector with RBAC to manage and query niche, domain-specific documents using the existing suite of agents
• Created MCP connectors to integrate SQL/NoSQL databases with injection guardrails, enabling agents to query datasets with millions of records via the web app or directly through messaging channels
• Developed a Task Board supporting one-time and recurring job scheduling for agents, with status tracking across the job lifecycle
• Luccro: Developed backend APIs and backtested custom trading algorithms on the Luccro Automated Trading platform, ensuring reliable end-to-end automation from signal generation to trade execution with a focus on higher win/loss ratios
• AxiomLabs: Contributed to a core AI chatbot system for construction management and risk analysis, leveraging data from internal emails, file exchanges, and Slack communications to identify delays and assess project risks; powered by a hybrid multi-tenant RAG system combining graph RAG and vector RAG
• Adassa AI: Developed a Table-Augmented Generation (TAG) system enabling users to analyze complex Excel datasets, automatically generating efficient SQL queries from natural language while minimizing mathematical errors and implementing fail-safe mechanisms against SQL injection

Research Engineer at DCube Tech, NSTP (2022-03 – 2024-06)

• Engineered efficient TensorFlow data pipelines for large-scale benchmarking on ImageNet (ILSVRC-2010)
• Conducted research on global attention-based deep neural network architectures for image classification, focusing on optimizing spatial feature learning with reduced parameter count and elevated learning rates
• Researched and developed a fish biomass estimation system from unconstrained underwater videos, integrating a custom-trained YOLOv8 model with Gaussian Mixture Models (GMM) and Optical Flow for robust fish detection and tracking
• Implemented SAM-based segmentation and monocular depth estimation models (MonoDepth, Depth Anything) to derive accurate volumetric biomass measurements
• Designed and developed a unified standalone network consolidating classification, segmentation, and biomass estimation into a single end-to-end pipeline

AI Engineer at Freelance Projects (2021-04 – 2025-11)

• Built an Image-to-Image translation system with CycleGANs on the GTA2CityScapes dataset
• Developed a music genre classification system using Mel-frequency cepstral coefficients (MFCCs)
• Re-wrote the chicksexer Python library in PyTorch to detect gender from names (textual data), training the model on the Milos Bejda Names Dataset of around 1 million words
• Built a human action recognition system using ConvLSTMs with a Streamlit GUI for real-time detection
• Benchmarked OpenAI GPT-4 and Meta Llama-3.3-70B on large-scale Arabic software compliance datasets, evaluating performance against Saudi Arabia (KSA) security and regulatory standards
• Trained Llama-3-8b using Llama Factory on an open-source Docker commands dataset

Bachelor of Engineering in Electrical Engineering – National University of Science and Technology (NUST) (2020-08 – 2024-06)