Jotaniya Keyur
@jotaniyakeyur
I build pretrained, fine-tuned, and aligned LLMs, synthetic datasets, and efficient vision models.
IndiaWhat I'm looking for
At BinaryCode Service, I designed and pretrained RoPERT-Base, a 108M-parameter BERT-style encoder trained from scratch on 10B+ English tokens for masked language modeling, classification, token classification, and extractive QA.
I've developed the Type-01 family of instruction-following and reasoning LLMs by adapting Qwen2.5, Llama-3.2, and Qwen2.5-Coder. My multi-stage training pipelines combine LoRA warm-up, full SFT, DPO, reasoning-focused LoRA, and GRPO alignment, supported by synthetic datasets for coding, STEM, translation, summarization, and tool calling.
I also build efficient vision systems, including ViT-Nano and modern lightweight CNN architectures, and an offline personal AI assistant with RAG, web search, controllable reasoning, and a quantized coding model.
Experience
Work history, roles, and key accomplishments
AI Engineer
BinaryCode Service
Mar 2026 - Jun 2026 (3 months)
Designed and pretrained a custom BERT-style encoder with 108M parameters, and developed fine-tuned LLMs and lightweight vision models.
Education
Degrees, certifications, and relevant coursework
G H Patel College of Engineering & Technology (GCET)
Bachelor of Technology, Internet of Things
2022 -
Grade: 7.13
Pursuing a Bachelor of Technology in Internet of Things with a CGPA of 7.13.
Tech stack
Software and tools used professionally
Availability
Location
IndiaAuthorized to work in
Job categories
Skills
Interested in hiring Jotaniya?
You can contact Jotaniya and 90k+ other talented remote workers on Himalayas.
Message JotaniyaGet matched with your dream remote job
Sign up now and join over 250,000+ remote workers who receive personalized job alerts, curated job matches, and more for free!
