AYUSH MAHAJAN
I’m Ayush Mahajan, an AI systems and product engineer who builds AI that has to work in real product conditions—latency, reliability, hardware limits, imperfect data, and operator control.
About
I’m Ayush Mahajan, an AI systems and product engineer who builds AI that has to work in real product conditions—latency, reliability, hardware limits, imperfect data, and operator control. My work spans software, embedded systems, and applied R&D, and I often take projects from architecture through implementation, including real-time voice AI, LLM-based evaluation systems, brain-computer interface experimentation, and the infrastructure needed to run them reliably.
I enjoy designing systems that fail predictably and stay safe in the field—whether that’s separating fast “live paths” from heavier analysis, engineering low-latency signal pipelines on constrained hardware, or designing deterministic communication layers that don’t become dependent on AI. I’ve led startups and teams, ran pilots with real users, and continue working on topics like world models, reinforcement learning, embodied AI, and AI red teaming for autonomous systems.
- 10 years' experience
- Joined November 2026
Skills· 16
Languages
- EnglishFluent
- HindiFluent
- JapaneseBeginner
- GermanBeginner
Industry experience
- Computers & Electronics
- Software & Internet
- Telecommunications
- Government
- Healthcare
- Professional Services
Work experience· 5 roles
Worked on AI and product-system concepts for tactical and off-grid communication products constrained by communication reliability, operator control, and predictable failure modes. Developed assistive/agentic intelligence architecture layered on deterministic communications so core communication behaviour is not dependent on AI. Considered intermittent connectivity, off-grid operation, and hardware/network constraints when shaping system behaviour for field deployment.
Built a voice-first founder-evaluation platform combining live speech capture, conversation state, LLM orchestration, evidence retrieval, scoring, objection handling, and investment-memo generation. After early testing, separated a lightweight live-call path from a heavier asynchronous analysis path to protect response latency. Ran 180+ sessions with 50+ founders and investors, turning feedback into scoring rubrics, objection handling, and report workflows. Led discovery, implementation, and pilot usage with a ~6-person team and stakeholders; product IP was acquired by a university-affiliated startup accelerator in 2025.
Founded and led a ~10-person team building Xone, a low-cost brain-computer-interface prototype, while serving as the principal engineer across electronics/PCB design, firmware, signal processing, and software. Developed low-latency communication and multithreaded Python processing for visual-attention EEG experiments on resource-constrained hardware. Secured INR 50 lakh+ in government-backed non-dilutive support and received STPI/TiECon recognition; presented to institutional stakeholders including LBSNAA.
Education
Certifications and awards
Hire AYUSH MAHAJAN today
Post a job and invite AYUSH, or contact them directly. No hidden commission.