Consider these 3 things when choosing your Machine Learning partner

The right ML partner will help you make predictions & recommendations across your business. This could be the most important decision you make this year.

Let me explain why I feel strongly about this topic. I’ve been selling for over 12 years with countless conversations, negotiations, experiencing highs and lows along the way. A good salesperson acts as an advisor caring as much for their client’s needs & desires as they do for their own needs and their companies needs. Let me give you more context.

When I was younger I just wanted to get the deal in and move to the next one, but after a while, that didn’t sit right. As my life experiences have grown and my career developed, I believe it’s my duty to help anyone I come into contact with, client or nonclient, deal or no deal. Sometimes help comes in the form of just shutting up, listening and empathising with the other person’s situation or introducing them to a different company that can help them.

I feel privileged to speak with many companies each week about their strategy and aspirations around using AI. Whether it’s removing weight out of manual processes, give analysts better insight or improve their customer’s’ journey.

Machine Learning (ML) is being used in a variety of ways, such as predicting financial risk and credit default, predicting supply chain demand, predicting where to deploy crops, to reduce energy consumption on electricity grids and even to automate cucumber farms in Japan.

Ram Charam, world renowned business advisor, state’s “Innovation must be the central driving force for any business that wants to grow and succeed in both the short and long terms”. Whichever way you look at it, Strategically, operationally, leveraging ML technology is a key part of your future success.

ML is happening now with innovation-minded business leaders embracing ML as “the next big thing”….allocating 15% of their IT budgets to ML, according to a recent MIT technology review.

With all this in mind surely this is the ‘perfect’ market for a guy selling machine learning capabilities? Much like that ‘perfect wave’ Patrick Swayze chased in Point Break or the ‘perfect storm’ the guys chase on the Discovery channel. Yes the conditions are prime, however, in some cases, buyers still seem reluctant to commit budgets to experimenting with ML DL technologies and suppliers. Here are some insights to help guide you through any uncertainties when taking the leap into your first ML project or moving from your first project to scaling ML across your business.

1. Choosing the right machine learning provider:

I’ve come from a background of keeping development in-house and I get the reasoning behind this, but with fast evolution of machine learning technology and shortage of data scientists, it’s vital to find an ML partner to fill this gap and get you going quickly. They can help you adopt predictive analytics as a core part of your business and provide infrastructure for you to become autonomous in running your own data science functions. An ML provider should help you identify and validate viable use cases across your business, guide your model factory through to deploying your models into production.

2. Consultancy vs Technology + Consultancy:

a) Consultancy

ML consultancy companies offer a proxy data science team to identify use cases and build / train your predictive or recommender models (having been granted access to your data in cloud or on-prem). There is a lot of value in this approach and is a necessary part of your ML discovery. However what happens when you want to move your predictive models from R&D into production? Let’s remember investments in data science and ML are ultimately made to improve key business KPI’s such as cash generation, net profit, process velocity thus adding business growth and value to your end users / customers.

b) Consultancy + Technology

There is value in working with an ML partner who’s also built their own open source technology. You can get a feel for their platform capabilities without committing budget or on a low level support plan. With confidence in their technology and credibility, you can discuss a partnership to benefit from their data science & engineering whilst leveraging their open source technology & seeking guidance on other tools they recommend. With the wide range of open source ML tools available, be wary of a ML suppliers that move you away from utilizing multiple tools to build models eg R Python TensorFlow Spark etc. If an ML partners technology is framework agnostic with no vendor lock you can feel reassured you won’t be tied down when your data science team want to explore new tools as they appear on the market.

3. Can your ML partner support your model deployment?

The companies that deploy the best models the fastest will win the race. Avoid your ML models being stuck in R&D by choosing a partner with deployment capabilities. You can prove a model works in a sandbox environment but you only improve your business KPIs & deliver ROI when making recommendations / predictions against live data. Choosing an ML partner that has the ability to quickly deploy machine intelligent solutions or who can provide you the tools to do this yourself when your consultancy contract is over is a bonus.

In summary, there is an ever expanding ML DL landscape with companies focussing on specific industry sectors or different parts of the machine learning pipe (infrastructure). It makes sense to work with ML providers with experience in your sector, but also consider if they are well placed to guide you right through the ML process past R&D into productionizing your models with no limitations on what model building tools you should use. With time to market being a key driver in winning the ML race, your ROI is realised when your models are deployed, which represents the gateway to enriching the lives of your companies valuable customers and consumers.

Join our closed beta

Do you want to gain early access to Seldon Deploy? We’re looking for active data science teams to provide feedback to help shape the product before launch.

To gain early access to the Seldon Deploy Closed Beta, please apply here.