Senior NLP Engineer

𝗦𝗲𝗻𝗶𝗼𝗿 𝗡𝗟𝗣 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿 - 𝗕𝘂𝗶𝗹𝗱 𝗩𝗼𝗶𝗰𝗲 𝗔𝗜 𝗧𝗵𝗮𝘁 𝗔𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗪𝗼𝗿𝗸𝘀

Remote | BDT 150k+/month

𝗧𝗵𝗲 𝗢𝗽𝗽𝗼𝗿𝘁𝘂𝗻𝗶𝘁𝘆

We're building voice AI that thousands of people actually use. Not demos. Not prototypes. Real systems handling real conversations, every single day.

We're looking for 3 Senior NLP Engineers who get excited about making AI sound human. You'll have serious hardware to work with (16x H100 GPUs and counting), meaningful problems to solve, and the freedom to ship solutions that matter.

𝗪𝗵𝘆 𝗧𝗵𝗶𝘀 𝗥𝗼𝗹𝗲 𝗘𝘅𝗶𝘀𝘁𝘀

Voice is becoming the primary way people interact with AI. While others are still figuring out chatbots, we're shipping production voice systems that handle everything from customer service to real-time translation. We need engineers who can take cutting-edge research and turn it into rock-solid products that work at scale.

𝗪𝗵𝗮𝘁 𝗬𝗼𝘂'𝗹𝗹 𝗔𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗗𝗼

𝗕𝘂𝗶𝗹𝗱 𝗣𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻 𝗔𝗜

• Design speech and language models that respond in milliseconds

• Connect STT, TTS, and LLM components into seamless experiences

• Tackle real challenges: multilingual support, speaker diarization, ultra-low latency

𝗦𝗵𝗶𝗽 𝘁𝗼 𝗠𝗶𝗹𝗹𝗶𝗼𝗻𝘀

• Your code will power conversations worldwide, not sit in a repo

• Build systems that scale elegantly under pressure

• Create data pipelines that actually work in production

𝗣𝘂𝘀𝗵 𝗕𝗼𝘂𝗻𝗱𝗮𝗿𝗶𝗲𝘀

• Publish research at major AI conferences

• Co-author patents that matter

• Experiment with our GPU cluster—if you can imagine it, build it

𝗧𝗲𝗰𝗵𝗻𝗶𝗰𝗮𝗹 𝗦𝗸𝗶𝗹𝗹𝘀 𝗪𝗲'𝗿𝗲 𝗟𝗼𝗼𝗸𝗶𝗻𝗴 𝗙𝗼𝗿

𝘋𝘰𝘯'𝘵 𝘩𝘢𝘷𝘦 𝘦𝘷𝘦𝘳𝘺𝘵𝘩𝘪𝘯𝘨? 𝘛𝘩𝘢𝘵'𝘴 𝘧𝘪𝘯𝘦. 𝘐𝘧 𝘺𝘰𝘶 𝘮𝘢𝘵𝘤𝘩 60% 𝘢𝘯𝘥 𝘭𝘰𝘷𝘦 𝘭𝘦𝘢𝘳𝘯𝘪𝘯𝘨, 𝘸𝘦 𝘸𝘢𝘯𝘵 𝘵𝘰 𝘵𝘢𝘭𝘬.

𝗖𝗼𝗿𝗲 𝗔𝗿𝗲𝗮𝘀:

𝗠𝗼𝗱𝗲𝗹 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲𝘀

• Experience with Transformers, Conformers, RNNT-Transducers

• Knowledge of CTC, TDT decoders, n-gram language models

• Voice activity detection, diarization, neural turn detection

𝗦𝗽𝗲𝗲𝗰𝗵 𝗧𝗲𝗰𝗵𝗻𝗼𝗹𝗼𝗴𝗶𝗲𝘀

• Spectrogram models, vocoders, audio codecs

• Speaker embeddings and audio tokenization

• Speech quality evaluation and benchmarking

𝗟𝗟𝗠 𝗘𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲

• Fine-tuning, RAG, agent/tool calling

• Token optimization, LoRA, multimodal models

• Production deployment with vLLM, TensorRT-LLM

𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴 𝗦𝗸𝗶𝗹𝗹𝘀

• Strong with HuggingFace Transformers, LangChain, or similar

• ONNX optimization for production

• End-to-end voice pipeline integration

𝗪𝗵𝗮𝘁 𝗪𝗲 𝗡𝗲𝗲𝗱

• 5+ years building NLP/ML systems

• Deep expertise in at least one area: STT, TTS, or LLMs

• Track record of shipping production systems

𝗛𝗼𝘄 𝗪𝗲 𝗪𝗼𝗿𝗸

𝗬𝗼𝘂'𝗹𝗹 𝗧𝗵𝗿𝗶𝘃𝗲 𝗛𝗲𝗿𝗲 𝗜𝗳 𝗬𝗼𝘂:

• Ship Fast and Learn Faster

• Launch MVPs in weeks, iterate based on real feedback

• Choose customer value over perfect architecture

• Fix problems in production rather than debate them in meetings

𝗢𝘄𝗻 𝗬𝗼𝘂𝗿 𝗪𝗼𝗿𝗸

• Start things and see them through

• Don't wait for permission to solve problems

• Learn from failures and share those lessons

𝗙𝗼𝗰𝘂𝘀 𝗼𝗻 𝗜𝗺𝗽𝗮𝗰𝘁

• Measure success in customer wins and revenue

• Build systems that improve themselves

• Know when to kill projects that aren't working

𝗧𝗵𝗶𝘀 𝗜𝘀𝗻'𝘁 𝗙𝗼𝗿 𝗬𝗼𝘂 𝗜𝗳 𝗬𝗼𝘂:

• Need consensus before taking action

• Can't ship until everything is perfect

• Prefer endless planning over rapid iteration

• Think certain tasks are "beneath you"

• Can't own your mistakes and improve

Read more about our core values here -

𝗪𝗵𝗮𝘁 𝗪𝗲 𝗢𝗳𝗳𝗲𝗿

𝗖𝗼𝗺𝗽𝗲𝗻𝘀𝗮𝘁𝗶𝗼𝗻

• BDT 150k+/month salary - Can be higher for highly skilled candidates. We pay for impact

• Two festival bonuses - 50% of gross salary each

• 38 days PTO - Because rest matters (14 paid + 14 sick + 10 casual)

• Any AI tools you want - Copilot, Cursor, whatever makes you productive

𝗧𝗵𝗲 𝗚𝗼𝗼𝗱 𝗦𝘁𝘂𝗳𝗳

• Unrestricted GPU access - 16x H100s today, more as needed

• Patent co-authorship - Your innovations, your name

• Conference presentations - Share your work at NeurIPS, ICML, ACL

• True remote work - No "hybrid" nonsense. Work from anywhere. We meet quarterly for team bonding.

𝗥𝗲𝗮𝗹 𝗜𝗺𝗽𝗮𝗰𝘁

• Build for global users, not just local markets

• See your code in production within weeks

• Work on problems that haven't been solved yet

𝗢𝘂𝗿 𝗜𝗻𝘁𝗲𝗿𝘃𝗶𝗲𝘄 𝗣𝗿𝗼𝗰𝗲𝘀𝘀

𝘕𝘰 𝘓𝘦𝘦𝘵𝘊𝘰𝘥𝘦. 𝘕𝘰 𝘵𝘳𝘪𝘷𝘪𝘢. 𝘑𝘶𝘴𝘵 𝘳𝘦𝘢𝘭 𝘦𝘯𝘨𝘪𝘯𝘦𝘦𝘳𝘪𝘯𝘨 𝘤𝘰𝘯𝘷𝘦𝘳𝘴𝘢𝘵𝘪𝘰𝘯𝘴.

𝟰 𝗦𝗶𝗺𝗽𝗹𝗲 𝗦𝘁𝗲𝗽𝘀:

1. Pre-test Call (30 min) - Meet 2 team members, discuss your experience

2. Technical Assessment - Practical take-home challenge (may skip for senior folks)

3. Technical Interview (75 min) - Deep dive with 3 team members

4. Culture Fit - Role-play and behavioral discussion

𝘙𝘦𝘢𝘥𝘺 𝘵𝘰 𝘚𝘩𝘪𝘱 𝘝𝘰𝘪𝘤𝘦 𝘈𝘐 𝘛𝘩𝘢𝘵 𝘔𝘢𝘵𝘵𝘦𝘳𝘴?

Apply now here →

Call for any queries

𝗔𝗯𝗼𝘂𝘁 𝗩𝗲𝗿𝗯𝗲𝘅.𝗮𝗶

We're building the infrastructure for natural AI conversations. Our voice AI systems are already powering thousands of interactions in Bangladesh and Japan. Backed by serious investors and growing fast, we're focused on one thing: making AI conversations so natural, people forget they're talking to a machine.