introducing - “the research lab”
Our research lab programme is set to commence in Mid-February and its first time opening our doors to our work, allowing AI enthusiasts, early-career professionals , graduates and undergraduates participate in a practical AI research programme.
Here are some key details about the programme and how you can participate in the next cohort.
why?
We want to bring together the curious minds that can’t stop thinking of new ways to improve things. We want the restless minds that sleep late and wake up early to continue working on their ideas. In short, we want people that tick the way we tick because we believe we can achieve more together. Team work makes the dream work!
Another reason we decided to create this research lab is to help tackle the unemployment crisis. Let’s take at the numbers for fresh graduates.
We can spend this whole article trying to figure the reason for these trends but we rather offer a solution. Instead of harping on the problem, we will attempt to provide some balance through our research lab.
Our hope is to give our researchers a playground to develop their AI expertise but also to develop the necessary skills to integrate into this constantly evolving workforce. With the slew of AI tools and the constant evolution of AI models, corporations feel less inclined to ‘bet’ on a young graduate in favour of trying to take advantage of the promised efficiencies of AI. Our programme, gives you the necessary skills to stand out amongst a sea of graduate postering their certifications.
details of the programme
Here is the first 4 weeks of the schedule:
Phase 1 - “The Collision”
Goal: Ingest the state-of-the-art and immediately break it.
Week 1 (Aggressive Intake): Rapid review of relevant research papers, articles, blogs and the Asycd knowledge base. The goal here is to obtain the necessary background to start drafting a prototype solution or a plan at least.
Week 2 (The "Crappy" First Draft): By Friday of Week 2, the researcher must have a "V0"—a script, a notebook, or a prompt chain that technically runs, even if the output is garbage. Milestone: The "V0 Burn." A demo where you show the system failing. You explain why it's failing based on the research you read in Week 1.
Phase 2 - “Pivot & Patch”
Week 3–4 (Iteration): Using the Week 2 failure to refine the architecture. This is the "Lab" phase—swapping out models, changing vector retrieval methods, or adding reasoning loops.
targeted practical AI experiments
We encourage applicants to select their curate their own experiments but we can always help mould an idea based on interests for both parties. Asycd has the following interests and we are always looking to increase our breadth:
AI Agents
Pioneering Long‑Term Memory: Designing agents that retain and evolve knowledge across interactions.
Self‑Evaluating Intelligence: Building agents that critique, correct, and refine their own reasoning in real time. This may involve creating multi-agent frameworks, fine-tuning using relevant examples or implementing artificial thinking through innovative logic patterns.
Cognitive Personalization: Crafting adaptive profiles that mirror human individuality and unlock bespoke collaboration.
Autonomous Research Engines: Deploying agents that independently generate, test, and publish new discoveries.
Software Engineering for AI
Next‑Gen Conversational Systems: Advancing voice assistants and chatbots into truly immersive, multimodal companions.
Closed‑Loop Innovation: Embedding feedback cycles that accelerate software evolution without human bottlenecks.
Intelligent Infrastructure: Pushing the boundaries of vector and cloud databases to power scalable, agentic ecosystems.
Our experiments are based on academic research but we are not trying to produce academic research ourselves. We are trying to create practical solutions to meaningful problems. Our research programme aims to instil an ethos of practicality through achievable and tangible AI experiments. Tangibility over intellectualism.
the 01 cohort
Saksham Kapoor - Distributed State Consistency & Cost-Aware Orchestration for Multi-Agent Systems.
Joel Allen-Caliste - Bayesian Prompt Compression via Structural Pattern Discovery
Syed Kumail Haider - Emergent Memory and Continual Reasoning in Generative AI Agents via Biologically Inspired Learning Dynamics
Victor Ibhafidon - Autonomous Debt Collection Voice Agent
