About yuvraj
I am a Machine Learning Engineer with expertise in LLMs, RAG, and scalable ML systems. With strong foundation in understanding Deep Learning Frameworks, I specialize in pretraining, fine-tuning, model-merging, evaluating and benchmarking LLMs as well as building powerful assistants for real world applications. Additionally, I also have hands-on experience of working with ASR models such as Whisper and HuBERT. Currently, I am exploring SOTA LLM models, Agentic workflows, and multi-modal LLMs to push the boundaries of AI capabilities. Alongside this, I stay actively engaged with the latest advancements in Machine Learning research, particularly in the field of LLMs and their evolving architectures. Let’s connect if you are working on LLMs, applied AI, or building the next generation of intelligent agents. Get in touch: GitHub (https://github.com/yuvrajsagar17) HuggingFace (https://huggingface.co/yuvraj17)
Key Skills
Experience
Llm Engineer
CurrentUpwork
* Finetuned Llama-3.1 8B, 70B, Llama3.2-3B, Qwen-2.5 14B and 72B on medical dataset using axolotl on runpod GPUs. * Merging adapter model with base model. * Inferencing LLMs using Transformers and vLLM engines. * Evaluated finetuned models on OpenMed and Nous Benchmarks. * Created GGUFs in 4,8 and 16 bit form for efficient inference on CPU and Apple devices using llama.cpp. * Carefully crafted and sanitized a medical dataset of alpaca format from latest datasets available on HuggingFace. * DARE, TIES, and SLERP merged different LLMs using Mergekit.
Research Intern
École De Technologie Supérieure
* Designed and implemented a multi-step code generation pipeline for DSLs generations using "vLLM" and "Outlines", integrating structured prompting, GPU inference, and automated output logging. * Built prototypes that convert NL into ready-to-integrate DSL configs. * Implemented structured outputs using "vllm" and "Outlines" for controlled text generation. * Developed context-free JSON grammar for the target DSL to improve generation accuracy. * Curated and standardized 50+ prompt–code pairs, building a robust dataset for model evaluation. * Created a custom evaluation metrics to measure the accuracy.
Education
National University Of Sciences And Technology (nust)
Bachelor Of Engineering
Interested in connecting with yuvraj?
Sign up for NinjaHire to send a connection request.
Common Questions
What is yuvraj's expertise?
yuvraj specializes in LLM Engineer - Remote, with expertise in agile modeling, algorithm analysis, analytical reasoning, analytical skills, benchmarking.
Where is yuvraj located?
yuvraj is based in islāmābād, jammu and kashmir, india.
How much experience does yuvraj have?
yuvraj has 2+ years of professional experience.
How can I contact yuvraj?
You can connect with yuvraj through NinjaHire by signing up for a free account.
Other LLM Engineer - Remotes
Looking for a different LLM Engineer - Remote?
Describe exactly who you need and NinjaHire will source them for you.
Type a role to try NinjaHire for free