Lokonova AI

Founder

Leloko James Lepolesa

Artificial Intelligence Engineer · Founder, Lokonova AI

Leloko James Lepolesa, founder of Lokonova AI

Artificial Intelligence Engineer with strong expertise in artificial intelligence and software engineering — bridging advanced research and practical deployment.

Industries

TelecommunicationsFacilities managementEnergy systemsHealth

Hobbies

ReadingSingingGym workoutsHikingSwimming
Chat with leloko.AI →

Leloko Lepolesa is an engineer with strong expertise in Artificial Intelligence, machine learning, optimization, and intelligent systems. He has working experience in Telecommunications and Facilities Management, research experience with contributions using deep learning and reinforcement learning to develop intelligent solutions for power systems, and short-term applied AI work in Geology, Climate research, and Health. In research, he has published in top IEEE journals and conferences and has reviewed for IEEE Transactions journals.

With a solid academic foundation and cross-industry experience, he delivers technically rigorous, scalable, and industry-ready solutions that balance performance, economic feasibility, and real-world constraints.

He is currently building SaaS AI products through Lokonova AI and providing consulting services to help businesses leverage AI to grow.

If you are seeking a solution that integrates intelligence with engineering — and brings both creativity and analytical depth to challenging projects — Leloko welcomes the opportunity to contribute.

  • Re-engineers learning algorithms under technical, economic, and human-centred constraints
  • Translates research-grade AI into robust, production-ready systems
  • Develops optimization and control frameworks for complex, interconnected networks
  • Collaborates across disciplines to deliver innovative, high-impact solutions

Education

PhD in Engineering (Electrical)

University of the Witwatersrand

July 2022 – July 2026

Completed.

Deep reinforcement learning technique for load balancing in power distribution networks in the presence of electric vehicles

MSc in Engineering (Electrical)

University of the Witwatersrand

Feb 2021 – June 2022

Distinction

Deep neural network classification for electricity theft detection in smart grids

BEng in Electronics

National University of Lesotho

Aug 2012 – May 2017

First-Class Honours

Awards:

  • Best Student in Electronics Engineering — Vodacom Lesotho
  • Best Student in Electronics — LEWA
  • Best Student in Electronics — Data-deal Security Systems

AI certifications

IBM AI Engineering Professional Certificate

  • Machine learning & deep learning with Python
  • Keras, PyTorch, TensorFlow
  • GPT, LLaMA, BERT, and custom models
  • RAG and LangChain for AI agents
Verify credential →

IBM Machine Learning Specialization

  • Regression, classification, and unsupervised learning
  • Recommender systems and collaborative filtering
  • Deep learning and reinforcement learning capstone
Verify credential →

Microsoft Certified: Azure AI Fundamentals

  • Generative AI, NLP, and computer vision on Azure
  • ML fundamentals and AI workload design
Verify credential →

Selected projects

Public-facing work and research. Private organisational projects are not listed.

Lokonova AI — live SaaS product

Dual Scope Digital

AI marketing tool for social media — industry and competitor monitoring, on-brand content, review, and publishing to LinkedIn, Facebook, and Instagram.

View project →

Lokonova AI — live SaaS product

Career Growth AI Copilot

AI tools for career growth — CV and ATS review, job applications, interview practice, and planning support for job seekers.

View project →

AI assistant

leloko.AI

Personal AI clone that answers questions about my work and collaboration opportunities, grounded in portfolio knowledge and powered by Gemini.

View project →

Recommender system

CV-to-Job Recommender

Extracts CV keywords, matches roles and locations, and surfaces relevant LinkedIn and Pnet postings in real time.

View project →

Research & engineering

Dynamic EV Charging Pricing (DDPG)

Collaborative deep reinforcement learning for load peak-shaving, valley-filling, and balancing across distribution networks with EVs.

View project →

MSc research

Electricity Theft Detection

Deep neural network classification with novel time- and frequency-domain features for smart grid theft detection.

View project →

Interested in collaboration, consulting, or a custom AI build?

Get in touch