Our client is a hyper growth start-up that's focussed on paraphrase generation (amongst many other things writing related)..

This company has 5m MAU and a team size of 40. The b2c writing and research platform leverages state-of-the-art NLP to deliver human-in-the-loop products which expedite and improve the quality of writing and research.

Their Founder reached to let us know that they need to find a Director of NLP Research who has worked on cutting edge NLG language models (GPT, seq2seq, T5, etc.), has translation expertise, deep learning and someone that can both train GPT-3 from scratch and know what's going on under the hood! For context they have already built an end to end co-writer that produces better more accurate results that GPT3…

Ideal Candidate:

• Doctorate/Masters degree in Computer Science or equivalent experience.
• 10+ years of experience leading software development projects with a distinguished track record on technically demanding projects.
• 5+ years of experience in NLP
• Solid understanding of the AI/ML domain and hands-on experience building new models and deploying them to production.
• Has worked with deep sequence to sequence models in the context of translation or similar tasks.
• Has worked with SOTA language models such as T5, GPT-3, and BERT
• Has a deep understanding of natural language generation
• Strong statistical analysis skills and demonstrated experience in deriving insights from unstructured data.
• Is able to think creatively about novel datasets and architectures.
• Good command over the English language.

Job Responsibilities:

• Lead a group of research scientists to solve previously un-tackled problems and advance the state-of-the-art in NLP
• Lead the execution of a high number of experiments in parallel, analyze results, and iterate.
• Lead the development and delivery of scalable transformer based models into production.
• Understand the product and collaborate with product teams to enhance the product roadmap.
• Bring thought leadership in architecture design and efficient research engineering processes.
• Proactively foresee issues and resolve them

Here's what the first 30/60/90 days will look like on paper but for complete transparency, they're looking to hire someone that can leverage their experience/expertise to help define this strategy for optimal success:

The first 30 days>

Get to know the team, the product, etc.

Understand team members strengths/weaknesses

Understand current state of the research roadmap and begin to make recommendations

Audit prior research and begin to make recommendations

Pick low-hanging fruit based on domain expertise

start designing experiment to train in-house XXL LM

60 Days>

Run end-to-end experiments

Take 1:1 meetings with direct reports, begin guiding their research direction/responsibilities

Proactively make adjustments to current research roadmap and guide company leadership on future possibilities (including new product features based on SOTA NLP).

Make recommendations on novel architectures

Begin the process of training in-house XXL LM

90 days>

Fully own the research team, taking responsibilities for deadlines and overall output

Help direct reports set their OKRs

Proactively mentor direct reports, meaningfully contribute to their growth

Start setting future research roadmap

Move the needle on core KPIs associated with model proficiencies on our various tasks (paraphrasing, summarization, GEC, language generations, etc.)

Get in touch for more information: lewis.ad@darwinrecruitment.com / +1 617 480 9327

Darwin Recruitment is acting as an Employment Agency in relation to this vacancy.