Question 1
Does a research contribution need to be production-ready to be valuable?
Researchers and founders often talk past each other here. What assumptions is each side making?
Deep Learning Indaba 2026
When Research Meets Reality: AI Engineering in Africa
This session invites researchers and founders to interrogate one of the most consequential — and underexplored — transitions in African AI: the journey from a research insight to a product that works in the wild. Three African AI practitioners deliver focused lightning talks, then the room takes over in structured open discussion.
Deep-Dive Discussion1 HourResearchers & FoundersNigeria · August 2026Sponsored by EQUAL Compute Network
Why Attend
The goal is not consensus but clarity: sharper mental models for how African researchers and founders can navigate the research-to-production gap together. The session contributes to African AI sovereignty by strengthening the feedback loop between research and application.
- Hear three candid practitioner perspectives on the research-to-product transition, including where conventional wisdom falls short in African contexts.
- Engage with opening questions designed to surface assumptions and expose productive disagreements between researchers and founders.
- Build new connections across the researcher–founder divide — two groups that attend Indaba but rarely have structured space to talk to each other.
- Leave with at least one concrete reframing of how you think about your own work, whether in the lab or the market.
Session Flow
Three provocative opening talks seed the conversation; the facilitator uses pre-prepared opening questions to guide discussion, but follows the energy of the room rather than a rigid script.
00:00 – 00:03
Welcome & framing — setting the question
Krupa Suchak
00:03 – 00:10
Lightning Talk 1: The thesis gap — when research logic and product logic diverge
Krupa Suchak
00:10 – 00:18
Lightning Talk 2: Reading to build — literature review as a founder's tool
Naira Abdou Mohamed
00:18 – 00:26
Lightning Talk 3: From pipelines to agents — what changes when you ship in Africa
Luc Okalobe
00:26 – 00:30
Opening questions introduced; room organises into brief pairs for 3 min
Facilitator
00:30 – 00:52
Open discussion — facilitated with opening questions
All
00:52 – 01:00
Synthesis, closing provocations & networking open
All
Opening Discussion Questions
These questions are designed to be genuinely contested — there is no right answer — and to draw out productive tension between researchers and founders in the room.
Question 2
When is it irresponsible to build a product from research that hasn't been peer-reviewed?
Especially in African contexts — low-resource languages, healthcare, finance — where are the ethical thresholds?
Question 3
What does the African market teach you that the research literature cannot?
What kinds of knowledge only emerge when you deploy in Lagos, Nairobi, or Dakar — and how do you feed that back into research?
Question 4
Is 'agentic AI' a genuine shift in how we build products, or a new label on old automation?
What does the answer mean for how African teams should invest their engineering time?
Question 5
Who is responsible for the gap between African AI research and African AI products?
Researchers? Founders? Funders? Governments? The Indaba itself?
Speakers
All three speakers are confirmed and attending physically. Krupa Suchak previously organised the Compute Workshop at DLI 2024 (20–30 participants) and DLI 2025 (50–60 participants).
The Thesis Gap
Krupa Suchak
Founder, DataSpires · Lead Organiser
Krupa's lightning talk will challenge the assumption that a strong research result naturally points toward a product. Drawing on DataSpires' experience building the AfriLink SDK, she will argue that the research thesis and the product thesis are different objects — and that confusing them is the most common reason African AI products stall between prototype and scale.
Reading to Build
Naira Abdou Mohamed
Head of R&D, ToumAI · Co-Organiser
Naira will make the case that systematic literature review is one of the most underused founder tools in African AI. His talk draws on ToumAI's pre-build research process for African text-to-speech, where engaging deeply with existing academic work revealed not just technical gaps but commercial ones.
From Pipelines to Agents
Luc Okalobe
CTO, Yamify · Co-Organiser
Luc will offer a practitioner's perspective on what the shift from ML pipelines to agentic workflows actually feels like from inside an African product team — not as a theoretical paradigm but as a series of concrete decisions about deployment in environments with variable connectivity and infrastructure constraints.
Audience & Theme Fit
Who should come
ML researchers (graduate students, postdocs, faculty) and AI founders or product leads — particularly those at the intersection of the two. No specific technical prerequisites; the session is designed for intellectual engagement rather than skill-building. Estimated attendance: 20–40 participants.
Sovereign Intelligence: Africa's Path in a Frontier AI World
The 2026 Indaba theme is fundamentally about agency: who shapes the AI that shapes Africa. This session addresses that question at its most practical — how African researchers and founders collaborate to ensure frontier research and products are built by and for African communities.
Logistics
- A room with moveable seating (ideally circular or cabaret-style) and a microphone for lightning talks. No AV or Wi-Fi required.
- Three-minute pair discussions before the open floor to lower the activation energy for quieter participants.
- Gretchen Adams and Kofi Yeboah of the Mozilla Foundation will moderate the open discussion. Speakers remain for networking after the formal close.
- This session complements a separate 3-hour workshop submitted independently and targeting students and practitioners.
Lead organiser: Krupa Suchak — DataSpires