Sardar Ali Maqsood

MS AI | Research Assistant at 7BOT | Expert in Edge AI, RAG & Time-Series Forecasting

Based in: Rawalpindi, PakistanMost recently: Independent Researcher, Self-Directed ResearchExperience: 4 years

Research Assistant and MS AI candidate with a background in Electrical Engineering, specializing in the intersection of AI and hardware. Proven track record in deploying high-accuracy (97-98%) models on edge devices (ESP32) and developing intelligent optimization systems that reduce energy costs by up to 40%. Experienced in building offline RAG pipelines with Llama 3.2 and conducting advanced research in time-series forecasting and low-resource NLP.

Experience

Independent Researcher

Self-Directed ResearchworkFeb 2025

Summary

Developing an intelligent home energy management system that combines forecasting and optimization for cost-aware scheduling. Designing time-series forecasting workflows for load and price prediction using LSTM and GRU-based models. Conducting low-resource NLP experiments, including transformer-based Akkadian-to-English translation. Building reproducible research pipelines in Python and PyTorch for preprocessing, training, evaluation, and ablation analysis.

Results

Reduced household electricity costs by 30-40% through intelligent appliance scheduling and peak-power optimization.
Achieved a high prediction accuracy with an R-squared (R²) score of 0.94 using LSTM-based forecasting models for load and price prediction.
Achieved a BLEU score of 38% on an Akkadian-to-English translation task using a dataset of 12,000 samples (8,000 training, 4,000 test) in a Kaggle competition.

PythonPyTorchLSTMGRUTransformersNLP

Research Assistant

7BOT Technologies – NSTPworkFeb 2024 – Mar 2025

Summary

Developed a smart health monitoring watch for elderly care with real-time sensor processing and alert logic. Designed an intelligent air purifier with automated air-quality detection and adaptive control. Deployed an object classification model on edge devices for real-time inference.

Results

Achieved 97-98% accuracy (within 2-3% of medical-grade appliances) for health monitoring metrics like BP and temperature through rigorous sensor calibration.
Successfully deployed classification models on ESP32 microcontrollers using infrared MEMS sensors for real-time health monitoring.

ESP32MEMS SensorsInfrared SensorsEdge DevicesObject ClassificationSensor Processing

Computer Science Teacher

CollegeworkMar 2023 – Feb 2024

Summary

Taught computing fundamentals, networking, software engineering, databases, and programming. Guided students through Python-based projects, basic data analysis, and problem-solving exercises.

Results

Successfully taught C, C++, and Python programming to intermediate-level students, adapting complex technical concepts for early learners.
Developed a personalized teaching methodology, tailoring explanations and project difficulty to individual student learning styles and paces.
Simplified abstract concepts like recursion and control structures (if-else, loops) through real-world analogies and iterative examples, helping students overcome common logic hurdles.

C++CPythonMentoringCurriculum Delivery

Skills

Technical

NLP

Time-Series Forecasting

Machine Learning

Python

PyTorch

NumPy

Pandas

Retrieval-Augmented Generation

Optimization

Agentic AI

Model Evaluation

Data Preprocessing

LLM Fine-Tuning

Scikit-learn

Prompt Engineering

Transformers

Matplotlib

Jupyter

Information Retrieval

Keras

Git

Semantic Search

TensorFlow

C++

Flask

Areas of expertise

Learning Speed

Experimentation

Implementation Ability

Languages

Urdu

English

Projects

Offline Secure RAG System

Lead Developerpersonal

Developed a fully offline Retrieval-Augmented Generation (RAG) pipeline using Llama 3.2 (1B) and LangChain for privacy-sensitive environments. Integrated ChromaDB for efficient document retrieval, ensuring high reliability and authenticity by grounding model responses in local data.

Llama 3.2 1BLangChainChromaDBPython

Education

MS in Artificial Intelligence

Air University, IslamabadArtificial Intelligence2024 – 2026

Thesis. Intelligent HEMS Optimization: Integrating Advanced Deep Learning Models with Adaptive Scheduling

Coursework

Machine LearningDeep LearningComputer VisionNLPLLMsRAG

BS in Electrical Engineering

COMSATS University IslamabadElectrical Engineering2018 – 2023

Coursework

Signal processingcontrolprogrammingembedded systemsapplied machine learning

Hire SardarReach out about a role, a contract or a conversation.For recruiters

Sardar's twin is AI, it can make mistakes.

More people on Kanvis

Browse more people Create your own Kanvis